".' ~' .

~'-

,,_

,,_

,''.;'

_" .. ,

', ... :

',- :;

'.' ~, '

.'

-.

.. '

_ .'

'~.~ .

.. ~ ..

', ....

,"

-~- ,~ ..

. ;

. . '

-, ...

. ~"

~, . ' ..

"

-': '"

.'

_,

,"'_

'. '

- .-

PARTICULARLY

.,.:_

Service

87185

~_ NATIONAl' TfiANSlAT!O,N-S CENTER _. ~

_ _ _ _~ __ .::Thl$.tr~-~~5tl~ is ~~ t;lrn'I~~: for

_ . :~::._j_;~;':~:".<.~( ,:_" priva~ USB ar~.~F~· iJaiy. It may'~~.i: .. ·:'__:·::'. _"-~'-'>. __ ~ nofbe:sold'or_,ubRsh;~ in any farm.' • - - ~. r: _. '- '_' .... ""'- without the permission of the ButhQr •.

"oC _-_ ... _~.:-}._~. - __ ~ ... '" _.

;:

.. -

'.

''t~

, '-';._.

CONTRIBUTIONS ON THE OPTICS OF TURBID MEDIA, PARTICULARLY COLLOIDAL METAL SOLUTIONS

Gustav Mie

Greifswald, Physikalisches Institut

Original title:

Beitrage zur Optik trUber Medien, speziell kolloidaler MetallOsungen

Translated from:

Annalen uer Physik, Series IV, ~, No.3 (1908) 377-445.

Translated by:

Patricia Newman

Technical Translation Service Sandia Laboratories Albuquerque, New Mexico

July, 1978

CONTRIBUTIONS ON THE OPTICS OF TURBID MEDIA, PARTICULARLY COLLOIDAL METAL SOLUTIONS

1. The various colorations of metals in the colloidal state have had different interpretations at different times. Formerly it was assumed that the metals in question (particularly silver) occurred in several differently colored modifications. A later theory was founded on the assumption that the colors were based on optical resonance. The principal proponent of this theory was Ehrenhaft [1]. Recen~ly Maxwell [2] demonstrated that the theory developed by Lorenz [3] for optically inhomogeneous media offers

an incontestable explanation for the colors of colloidal metals if the suspended metal particles are very small. For the case of a fine metal suspension in which the particles are very small with respect to the wave lengths and the distance between them, the theory gives a very definite absorption curve from which the optical constants of the metal can be calculated. Although it is very different

from the absorption curve of the solid metal, it nevertheless has nothing to do with resonance in the sense in which

this term was used by Ehrenhaft, Wood and others. Thus Maxwell-Garnett, for instance, freely derived from Lorenz' theory the red color of many g6ld solutions which Ehrerthaft

regularly, predominantly in octahedral form, it is nevertheless very possible that upon rapid precipitation from the solution very severely deformed crystals may form and be suspended in the fluid as flakes or rods. On the other

hand however, regular crystals may certainly form, as observed, for example,' in the beautiful copper octahedrons in the so-called Avanturin glass. In theory it seems permissible

to substitute spheres for homogeneously developed crystals

such as octahedrons, cubes etc., and flattened or elongated ellipsoids for the flakes and rods. I however would like to mention that the observations made by Mr. Steubing on the polarization of the diffusely scattered light, which I shall refer to again in § 20, suggest that we are really not

dealing with spheres or similarly structured bodies.

The present work deals only with the simplest case, in which spherical form may be ascribed to the particles.

Polar Coordinates

2. For the sake of simplicity we shall number the coordinate axes (x, y, z) of a rectangular system of coordinates (1, 2, 3). Let r be the radius vector of a point (x, y, z), let abe the angle formed by r with axis 1,

and $ be the angle made by the projection of r on the (2, 3)plane

-4-

with axis 2 (Fig. 1). Then (r, 8, ~) are the polar COOT-

dinates of the point. Furthermore we want to call the

components of the electrical and magnetic field intensities

in this polar coordinate system.

Then it can be shown that Maxwell's equations take the following form:

2.

rl. sin ,'} (A. ,a (f~ + A. <t) ... ~(r .• in 8 . .$).,) _ aCr • .1:1,,) I

at r a~ a,p-

r. sin o, (It. i I!~ + .I/. (fa) =~_,~, __ iJ (r, ~in :'. ·f>:I_~ •

at cJ'p ar

r. (It. 8 e, + .1. (f ) = _~_(:::~~~ __ a;~.._,

at '1' ar 0:1

(1)

2 • "- a '\\'

-" . sm ,,. . p. 8 t

• (J 8,()01

- r . 8111 , ' • po. 8 1-

8.()." -r""'iil

_ 8 (r. sin 8. (!",) a (I'. (i~l

---'-'-~--- -- - -a-;;:--'

_ 8 (f, a Cr .• in ", (!.,)

- - 8.,,- - ----iJ;'--'

eHr. (!,11 a <2.

8" - - at,"'

Fig. 1. Coordinate system

Here k, A, ~ stand for die1ectricity, conductivity and

permeability in the same system. In a nonconducting medium,

2 .

k·~ + l/v , where v is the velocity of electromagnetic

waves in the medium.

By elimination we can arrive at a second order differ~

ential equation for (fr and ~r that does not contain the other unknowns. This will be done however, only" after we have transformed the equations for the problem of regular

oscillations.

-5-

We set:

(i - 13 t'~ni.' c:~ - 11 .2 .... in' etc.

,. ,.. "\,- r''''

(2)

where E and H only depend on the coordinates and not on

l' l'

time, and n is the number of oscillations per second.

Furthermore:

(3)

and here A indicates the wave lengths of the oscillation in

question in vacuum, and m is the complex refraction exponent

of the medium for light of wavp. length X. In a nonabsorbing medium (e.g. water), m is identical with the usual index of

refraction, but in metals 711 =" .(1 - i.;t) , where v is the real

quantity conventionally called the index of refraction and X is the absorption coefficient.

Finally we introduce the following notation:

_ .!.:..,!.,__,u_._l • H = ..1/

m r r

etc.

(4) (5)

2"m,. -l-=z.

We now arrive at the following equations in which the quantities E and M occur in the same way:

-6-

r' • sin 1" . E = iJ (% • a,in ~ . J~2 __ ~ (~. :'l{})

r a~ ocp'

r. sin o . E" = ,iJ ~~~, _iJ (x si~_~: .1f.P~,

iJcp ax

e (7' E.?) U '/,

o Ix . sin ,'t • }_'q.)

_ -----~.----- .. - .

az

(6)

r . E", = ~J~:!!'?2_

Z2. sin {}. JV = ~X_S!Il!.: E",)

r au

• ~ iJE

Z • Sin 1 • .11" = iJ cpr

s , _II.,. _ ~ ~-_._ ~,)_ iJx

Now we consider a spherical particle of radius p whose

center coincides with the origin and which is surrounded

by the solvent (e.g, water), The fields in the sphere and

solvent will be designated with the suffixes i and a respectively, the two indices of refraction will be m and

mo' We assume permeability to be the same inside and outside:

lJ· = 1..

lla'

The variable x has a discontinuity at the surface

of the sphere, since

where A' indicates the wavelength in the solvent. The following boundary conditions must hold at the surface of the sphere:

E6" - Iff"" A'.,. = s.; \

(.1'. Mu) .. '_ (or. ],[")if (or. J1/",). = (.1' • .11",),.

(7)

- 7-

Solutions of Maxwell's Equations

3. The methods developed by Rayleigh "Theory of Sound"

are used to solve Eq. (6). This has already been done by

Hasenohrl [6] for other purposes. In order to have together all the equations required for the discussion, I shall give below a brief derivation of the integrals.

As already mentioned, appropriate eliminations from

Eq. (6) provides a second order differential equation for Er and Mr:

_a!_(:z;~~!~ + =r _1 _, ,_i1_ (sin o . _iJ~~)

a Zi 810 ,'I a ti a :J

, + _.} ... ~tA__ + r2 Ji) = 0

BlDt tr a rp' ,.'

~~~.~~ + .-;-! _ .. a _. (Sill o . !:_}_!~.)

a Zl 8111 .'1 a:J a {I

+~-!-- .. a~,Jl:~ + .1,2. J[ _ O.

Bill' ti a.,,· ,.

(8)

We can now divide all solutions of Eq. (6) into three

groups. The first group represents the waves due to elec-

trical oscillations of the sphere~ it is characterized by:

E,. =F 0, JJ[,. - o.

The second group represents the waves due to magnetic oscillations of the sphere. For it we have:

Er=O, M,. =F 0.

-8-

The third group contains all integrals of the Maxwell

equations that represent regular periodic oscillations.

They can be found by adding the integrals of group I to the integrals of group II.

Assuming that we found a solution of the differential

equation for E , we find the remaining components corresponding

r

to group I very simply by the following procedure: we

substitute Mr = a into the second and third equations of system (6) and then substitute for M~ and Me the values

from the fifth and sixth equation. We find immediately equations which can be used to calctilate Ee and E~ from the known Er. Now if we have Er, Ee, E~, we naturally find Me and M~ from the two last equations of system (6). This gives us the following system of equations:

it, ... 0, 1

() (Xi ~1I.,) = r . s; ,

_~(~ M",l .... _ z . E6 •

c:&

(9)

The calculation is the same for group II. Furthermore

we can calculate E in general as a sum of terms each of r

which fulfills Eq. (8) individually and is a product of a function of ~ with the function of the angle 9,~. The

'Vth term is:

ECr) K.(:c) p tl.

.... - _._-. ('f" rr.)

r :1:' ,,- , ..,~ •

-9-

Then K and P must satisfy the following two equations:

\I \I

II' i; (1 rp) to

-_- + - - - i\ = 0

.i 2;1 :r:". I

1 a [. t o P ] 1 0' P .

--. --.. sIn l'f. _. • .•

liu:J iJ:J 0 .'1 + .iti' ~'t •. 8q," + c". P" = 0,

(10) (11 )

where a may be any either real or complex number. .\1

want to use functions of K and P for which

\I \I

We now

c. = II. (JI + 1).

(12)

Then P is a spherical function of variables 8, ~ of whole \I

number order \I, K is a function closely related to the \I

cylindrical function with broken index (denominator 2).

Equations (9) now give the following solution of Maxwell's

equations:

ECp) = K. (x) • P ('t' )

r . .xt • I ,(f ,

(13)

E:;' .... __ 1 __ • K.:_!x) .O/!!,_

".(v + 1) x 0 & •

BC.) _ 1 K: (x) 0 I'.

" - ... (, +1) . Zosln' -.1·13;';- ,

~t.I··l _ 0

JU,. - ,

.11:''') '"" _.! . ~_~,,_(x)_. ~I!".

".C .. + 1) z.sin {J 0 rp

Jr;l .... 1 . K,._~~!.. _~l!.!'. ..

". (r + 1) z a ti

Thus we have found the solutions of the first group

(Mr = 0). But now because of the symmetry of system (6) with respect to the E and M, we can immediately write the solutions

of group II (E = 0): r

-10-

.t'~~) = 0,

E!;') = 1 • _K .. (xL •. 0 \I!'__(&, ",) ,

JI.(I'+I) z.sin:t ocp

E(") = _ . __ .1_. __ • _.E~S::L. ~_~.(~! p~

., I' • (I' + l):z; iJ tJ- '

JIM K. (x) m (0 )

•• = -;:,-' +'. " (P ,

.11~') = _.1 _. . .. K:_~:r.l. a ~:. ,

".( .. + 1) :z; 8&

Jl~) :=II' .1 .• ~~: (x) _ • ~.~'!. ,

,.. (,. + 1) x .ain tJ- a cp

(14 )

where Kv and ~, are any solutions of Eq. (10) and (11) in which we can set c = v.(v + 1).

Now we must express a plane wave, namely that which

strikes the spherical particle, with the aid of (13) and (14). But first we must know the most important properties

of the solutions of (10) and (11).

The Functions K and I

4. Differential equation (10)

can be solved immediately for the case v = O. Two particu~ar integrals are:

Ko (.1') .. ,'.,

Ko(-r) = «r!=,

-11-

Furthermore, if we know the vth solution of the equation, X , then by substitution we can easily demonstrate that the

v

(v + l)th solution of the equation, Xv+l' can be calculated

as follows:

K +1 (x) = i. ((I' + 1). K. _ ~1_K_!_).

• Z d z

(15)

If we chose XO(x)= eix, this recursion equation gives:

K (x) :::: t: i.. );. _(~_!/'1_! - . --~"--. (16)

• • ~ ,,!(I'-:,,/')! 2" .zI'

_ 0

Differentiating once and eliminating dX Idx and d2X Idx2

v v

with the aid of (10), we find from (15):

.....

K (' ') .> .' •• ~."( •• i,,~. d 11.)

"-1 r = - I. 1'. - + --- . z dz

(17)

From Eq. (IS) and (17) we find the following often used

relationships:

(2 ,,+ 1). X .. (z) == i , K + ~ . K 1

' z· .-1. , .+1

(2" + 1). dol!" == (" + 1). i. K._l - V"~-' K .. +,.

(18)

A second particular solution of (10) is found immediately, since in (10) we have only the second power of x and dx, if we reverse the sign of X (-x) in the first solution. The

v

gener a1 solution' is thus:

A . K,. (r) + B. K,. (- :1') •

Since in the following discussion we will use the

-12-

solution K (-x) immediately, it IS appropriate to write v

the relationship that we obtain by reversing the sign of

x in (15), (16), (17), (18):

p

K (- .2:) = e> iJ:. ~ •. (~.~J') ! ... J.:- I)" I

y Lr ,II !(v-I')! ~/' .J;"

U

(19)

Kp+J (- ~.) = - i. ((11 + 1) =:: - X,: ( - x)) I )

K () . ( K. ( - z) l' , ))

r-l - x = + I. 11· -- -:zi - .. + .A" (- X •

(20)

(2 1)' K" (- x) .~' ( ) 1 I' ( ) .' 'J

-11+ '--:z;-=1 •. 11.-1,.-.'1: +-{'l1PT1 -x,

- (211+ I).K,,' (- x) = i.(JI+ 1). K,,_l (-x) - ;' .. Kr+l (- z) •

(21)_

s.

ix

If in expressions (16) and (19), we expand e and

e-ixin power series and then multiply and arrange in order

of powers of x, we obtain the power series expansions for

KoJ (x) and K/ -x), which naturally contain a finite number

of powers with negative, whole exponents.

We find:

r_ +00 •

K (x) = y..i'·.r". ~ __ ._(~_!-J~.!(_:-_~t' ---.

p _' ..,:_' (,I + r)! ,II! (v - p)! 2'"

r--r U

Now, however, it can be shown that:

This expression is remarkable in that it is always

different from zero and equal to zero in turn up to l' = V •

v -v+211 ..

Up to the power x only terms of the form x ·remaln In

the series. If we substitute l' = - V + 211 into the coefficients

-13-

we obtain:

(r - ~ + I) . (r - ,. + S) .• (r + ,. - 1) -·-------(ri- !')!--- --- -------

Thus:

(22)

r-+oo

+ ~(r-"+1)(r-,,+3) .. (I·+"-I) .

/ r • ,r z'

~ (r+,,)! •• •

r •• +l .

Likewise:

K.(-r)

(23)

We see that differential equation (10) also has one

solution which can be developed around the zero point into

a power series without negative powers. We will represent

it by I and set: "

(24 )

The power series for I has only real coefficients: v

(25)

-14-

From Eqs. (18) and (19) we find two important relationships:

,

(2v + 1). 1. = J + J .)

';1: .-.} .+) I

dl

(2 11 + 1). -...!.. = (II + l}, J - v. J

Ii:r; .-1.+1 •

(26)

6. To calculate the functions K (-xl, I (xl and their

v v

first derivatives (which alone appear in the final equations),

we use the following expressions:

(28)

(27) K.( - x) = (- W :.. .. 3 :.:~: ~~_1!. e : i&

! ~(;.!2

. (1 + ~ . .J~_::.~l~=-I_t __ . __ . ;c20 )

~ (2,,-1)(2,,-3) .. (21'-211+1) 1.3 •• (2 a - 1)

~+l) _ 1

+ix.(l+ ~~_ .. ("-:(f_~1),,(_-:1)" ...• _. ~2~"_)\1'

+ (2,,-1)(2 .. -3) .. (2,·-2a+1) 1.3 •. (2a+1)

(J(I'(-x)=+ :r;: .e-i -. ((1-x2)+ix),

K;(-x)=+"i~ .. riz.((l- }x2)+ix.(t- ~x2))1

K.' ( - x) = - ~5. ~ . e : I z. ((1 __ !. x2 +L x') + j x . (1 _ _!_ •. \'1)) .

~ x' 16 45 16

( K) (- xl = - .: . e : i",. (1 + i x) I Kz(-x)=- :' .ri:l.((I_ ~ x2) +i.x),

K3 (- xl = + _l~;i .. e : iz. (( 1 - : x2) + t. x , (1 - /5 .1'2)) I

. . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . ....

X· ,. , ( ) _ ( .). 1. S .. (2 I' - 1) _ i '"

• - x - - - I • 'II' :t'+ 1 ..• e

! . (+1)

"-;,-- 2 a-I .

• (to - IT)., + -- .. _- . (JI - U)"_I 2.

. 1 a -1 II • x.

( + 4( )"(2to-I)(211-3) .. (2I1-2<r+l) l.S .. (:!11-1})

r(-l)

---. ",,+1 l

2 (II-a-l).,+---.(,,-cr-1l.'-1 20)

+ i. s , (1 + ~(-1)".-.--.---- " ._ -. " ._ .. --~. ____;--I'

..._. (2 II - 1)(2" - 3) ••• (:! II - 2 " + I, 1.3., (2 a + I)

1 .

-15-

1 ( ) _ x' (1 3 x~ 3 x" a ' . ~'.. " ,).

lZ- 3'.-~'3!+1'bT-9·-"(+""

l.(z) = .:c~, (1 _ 3 , .. x~ + ~,_~ _~.:..5_.~~. + )

.. 15 7 3! 7.9 5! 9, 11 7! ... I

1 ( ) x, ( . 3 x~ 3 . 5 x' a . :; . 7 x· )

(29)3 ;r = 'l(i;,' 1 - '~.' 3! + 9. 11 . :;! - O~ ii:-13 . 7T + ... ,

.. .. .. " . .. .. .. .. .. .. .. .. . " .. .. .. ..

(SO)

1)' (:r) = ~~. (1 - 2 . ~, x' + 3, ~ . ,'t' _ 4. ~. za + )

, :I [) a! 7 5! !I 7! .• , I

J:'(z) = \5~' (1-

13' (~,) = 4. XS • (1 _ 104

a Xl 5 x' a x' )

7'31+ 9 "5i -Tl"7!+' ".' ,

3 8 x' 4 3, f) Xl a' s' a! + 2 '9'~-il' 5f

I) a.r,.7 x' )

-2 . 9:11:1'3 . 7T + . .. I

l'(,r) = _JI~+ !2:~':'_ .(1- ~_~ ... ~,.,' x'

.. 1 .3 .. (21' + 1) I' + t 2" + 3 3!

',.+5 a.5 x· )

+ ;+1' (2" + th01 " + 5) 'K f - . .. .

These series are much more adapted to numerical calculations than the finite expressions obtained for Iv and I~

from (24).

For other purposes however, we must also know

them:

I (Iin:z:

I x) - - cos x + --, . x

(2911)

1 ( ) . 8. COI:Z: S • sin:z:

2 z = - sm:r - -- + -_._,

:z: x'

I() 6. Bin:z: I5.cos:z; Ill.aina:

3 :r = + COSZ - -- - --_ .. ,.- + -.

a: :z;1 XS

.. .. .. .. .. .. .. ..

..

I ( ) . ( ,. 1J ) "". ( II· - rJ ")

.. :r = SID Z - -2- + .L,." Sill x - -01-

1

(I'+")! 1

. (;-_ r) TrT' 2 r.:If •

-16 -

I, ( . COB:Z: sin :Il

1 or) = + SID or + -;-- - --xi -- ,

I:' ( ) + 3. sin::r; + 6. cos :z: 6 • sin ::r;

z or - - cos or -::r;- -x'- - ---;.-,

'3'(Z) = _ sinr _ 6.cos:z: + 21.8~1I::r; + 41i.cos::r; _ 46.sin:z:

::llIa} I . . . . . . .::r;. • , :z:,~ , • .: , • ::r;: "

~+l

I J: (r) - COS (r - .'_;~) + 4cOS (r - (,: -t")

I ( .. + r - I)! ( .. (t. + 1) + r(r - 1»

·( .. -r+l)!r!·----~r--·

The Spherical Functions

7. The conventional spherical functions, which only depend on one variable (the zonal spherical functions) cannot

be used for this problem. We shall see in the next section that functions of the following two forms are required for .t he solution:

P, ({}, (p) = fl. (v), cos {} I

~, (t'J-, tt) == fl. (v). sill ()-, cos (P , V = sin 11- • sin cp ,

(31) (32) (33)

The function IT must, as we can see by substituting the '"

expressions for P", and ~. into (11), belong to the following differential equation:

d'

"(T;t((l - v2). u.) + ",(21 + I), ll. = O.

(34)

The geometrical significance of the three angle

functions found in (31), (32), and (33),:

- 17-

cos {} I lin {} . sin (I, sin {} . cos rp

is immediately evident. If we call the angles formed by the

radius vector of the point under discussion with the axes 1, 2, 3 (cf. 2) 81, 82, 93, then as we can see immediately

I.

cos ,'J-1 = COli ,'J-,

cos {}, = sin t', . COli (I,

cos {}3 = sin {}. ain (I ... fl.

(35)

from Fig. 2:

Now we want to collect the most

important equations for functions

of IT. Of course one particular \l

solution of the differential

equation of the spherical function

2.

Fig. 2. Trigonometric relationships

is rational and whole in the trigonometric functions of the angular

coordinates e,.. Only this solution is of interest.

If we have found it for one index (e.g. \l = 1), we can calculate for all other indices with the aid of the

fOllowing recursion equation

(36)

This equation can be verified immediately by substitution·

into (34). A solution of (34) for the case \l = 1 is then:

ill (II) = 1.

(37)

Then using (36) we find:

-18-

Il 35 3 15

,= 2 v - 2 v,

!:..I..:!...!.!. -1

.)

- " 1

n __ L. ")~(_1),.(2~=-~~!._.V·--'- .

.. - I,' .-- ( .. - al!'! ( .. - 28 - I)!

• 0

(38)

flo (Il) = 0 ~ ill (t·) = 1, fls"'" Sv, n=~~'V2_.!.)

3 2 2

Another series expansion for IT is the following: v

.. (+ 1)._1 a

~ " (I'+11! tI,,-2,-I.(1_r','

0- ,(- 1,. . .. -----.--.-.

,. " ", + 1

. '0 .!(s + I)! ( .. - 2. - 1)! 2·

(39)

For the special value v = 0, Eq. (38) gives:

0.(0) =- 0, when v is even,

.-1 (40)

[1 .. (0) - (- 1) ~ ( .. + n: h . dd

. '-;r e-.-t ..!.)~~~~)~. , w en v 1 so,

and for the first derivative:

""hen v is even,

(41)

when v is odd.

,.

Similarly from (39) we can calculate the values of

IT and its derivatives for v = 1. v

-19

8. Just as for the above mentioned functions Xv and Iv' we also have for ITv besides the recursion equation (36) a second equation that can be demonstrated as was (36):

~- = v.!!..!!.:.- (II - 1). fl •

"" d » "

(42)

From (36) and (42) we obtain the two often very useful

equations:

(2 If + I). a, "'" ~ !!,._t-! _ d ll._!_ ,

d » d » I

(2 I' + 1). v. rIr= 'II. fl.+1 + (II + 1). fl._1•

(43)

9. Finally the differential quotients of Pv and ~. appearing in Eqs. (13) and (14) shuu1d be noted here.

r, = tt •. cos .'t I 1

or.. d(r.lJ.l". EL+ d Il, ..

-0 -11- = - - -- riD---- • 8J1l " -a ,; -. SID 'P ,

-:-l_ . ~ I'. = ~.!!~ . cos ,')-. cos (f •

alii II iJ 'P d .,

~~r = tt •. sin o . cos rp , 1

o 'Ur d (" . 11.) ...

iJ tJ- _::& - -((,- _ •• cos II. cos rp .

__ 1_ • 0 '~r ... _ d (II, lr,,) sin + d 11. sin '}

sin.It a 'I' dr' If d " - • ,.

(44)

(45)

In particular for the first two ordinal numbers we have:

PI = cos tr , o!\ = _ sin l} iJtJ '

_ _!_ __ • _o?j_ = 0

sill II 0 'P ,

~I = sin t'J- • cos rr , o~

-8-:i- = cos ,?-. cos (f I

1 0 'l.!, •

-. -- . -- =- - sm (f

em o 0 rp ,

P, = 3. sin ,,'j-. cos {)-. sin 'I'»

iJ P, 3 ')?-'

'iFf'- == • cos -, . Sill rp I

1 it 1\!J I, --

SIii-~ '-a ~p- = v • cos II' • COS rp ,

~2 = 3. sin2l)-. sin (f. cos,,'.

o'l!, 3 . "n. .

-8 tt - = . sm :.. IT. sin 'P • cos (f ,

1 a 'U. 3 " 2

-, _, • --" - • Sl n t'J' • COli "'.

sm II a '1' 'r

-20-

InteSral Values of the Spherical Functions

10. Below we must calculate the total radiation reflected

by the sphere. We shall see that this problem reduces to finding the contour integrals

~"I 2."1

J JPr • P" . sin ,')- . rl o . d (r-

u u

and

:. 2 ...

JJ~~. ' tl,• ' sin .'f , d {} , d rp

U II

These integrals can be calculated very easily, for example by a method used in Maxwell's Treatise on El~ctricity [7]. We find:

:r :!." :t ~:,

fJp., P" .sin {J .tUt,(Ir, .. fJ~r' ~,., sin o. d o. d 'f - 0,

uu uo

(46)

:t ~:r :I ':!n

JJP:. sin .<J-. d o . d 'f = fI~::. sin ,'}., d ,'I-. d 'I' = 2n· I';~ .. : {l.

uu DU

Plane Waves

11. The present problem can be considered solved as

soon as the plane waves incident on the sphere are resolved into expressions of the form of (13) and (14).

We shall chose axis 3~ in the negative direction~ as the direction of propagation of the plane waves. We assume the

light to be rectilinearly poZarized~ and axis 1 is the

direction of the electricaZ oscillation~ thus axis 2 is the

direction of the magnetic oscillation.

-21-

If we again call z the coordinate of a point in the direction of axis 3, then according to (35):

z = r . cos N3 = r. tI •

The plane wave is thus represented by the following quantities:

') :!:I i=

tr- ... .. " i .. I + 1.' re tc 0

~I' • \t: = \.:.3 == ,

/,,:- ~:rir

~, ... _ I .z.. ~.~ i " It" ~- 0

.... , ., 3....·

- . II

Here k and ~ are the d1electricity and permeability in

the sense used in (1), and we have assumed the conductivity of the medium through which the beam passes (the water) to

be zero, AI is the wave length in the medium (water). Now

if we set ( c f . ( 5 ) )

2ni;t 2ni.r.1I

-.l'- = - .... ,- .. - = i.r.v,

then the quantities E , ••. and M , ... , which were introduced

, r r

earlier (Eq. (2) and (4)) can immediately be written as:

J.: - e;_~··. cos {}, £., .. - cl ••• ~. sin {},

J.~- 0,

.:11,. == i . t;' III. v. sin .'t . cos cp , .1ft)= i.ei.".v.cos{}.coslP, A[., "" -' i , e I." .•. sin rp •

(47)

But now the exponential function. can be easily developed into I and IT functions. We have:

._00

tlu == ~ (2 II + 1). ir-I. _!I::~r) . n, (v).

.-1

(48)

-22-

The validity of this equation can be most easily demonstrated by differentiating with respect to x and v. If we

I

first call the series f (x, v), then according to (26) and

(43) :

go

.~f_ = _!t-' { ~j.-I.(2'" + 1). (rtf" - 2.iL). tt (v)},

a·x :t L·r cl . .t::t P

• 1

00

c= '~'(' 2J ((II -1). j.-2. 1,.-1 + (II + 2). i' .1.+1), 11. (v),

1

00

• ~" 1 I. . tt. ( I) tt )

= I. "L./I - .-x:; .. (II .• +1 + II + .. -1.

1

00

. ":(2 I) ',-·1 t, II (

= l • V • ...:::.: II + I . "I' • V),

1

= i. v • ((r, v) .

Likewise we can show that:

a f . t'( )

7[; = I. X • 3,', v .

And firially:

..... _

Thus

{(x, v) = fliu.

Accordingly we have the desired series expans i.on.vfo r E r

and M : r

00

'" ,-, 1

D = >r(:! I' + 1)', ir:". __ •. P

, .- ,l;~ .. '

I

00

JfT = ~ (2 " + 1). j r . _!r .. In

r L~ x. 1-"',,'

1

-23-

where Pv and !:IS. are the functions defined in (31) and (32).

Now from the values found for Ep using (13) we can find a solution for group I. Furthermore using Mp and (14) we can find a solution of group II; the sum of these two

solutions must then be that integral of the Maxwell equations that gives the plane wave. We can easily demonstrate by

direct calculations that the resulting series expansions

are in fact identical with expressions (47).

Then from (26) and (43), keeping in mind the differential equation (10)

_. __ .~ . __ . ("'1. + I ) = Jr

r , (I' + 1) d .&~ • Xl

we can derive the following two equations

00

~ 21-+ 1 (. d(vHI dl in)

-' p._ "'_" __ ir:". T • __ ... _r_ + [r, _p • _I __ !:. = ()

- ,.. (I' + 1) .. tl v Ii z J " '

,

""

= :St:!l' + 1).i'·-I. I •. tlTI,.,

- ,1;' d »

I

(49)

00

~ _?~ I .. (;>-1 . I .. t! n. + i>. cl 1 z , d (v: 11.:,)

- ",(,'+1) ,. d n d » d »

I . .

....

= )!(2v+ 1).i •. 1 •. rr

..;...; X·I

1

If then according to (13) and (14) we form for

example:

-24-

then using (44), (45) and (49) we find:

E., -= - e i u . sin ,?- •

Likewise:

00

E., _ ~ .~_,,_ + I_. j,-I. _~!_ •. ~ 1)!_

_' ,', (" + 1) x • Bin & a '1'

1 00

- '5"! '!!_-i:"_I_ . t» . ~. ~L _ O •

...:..: ".(1'+1) x a~

1

The procedure is similar for Me and M~. Thus for the plane wave'. instead of (47), we can use the following:

eo

E .. ~(2" + 1).i'-1.1 •.. p,

r ,L.;' %:1 r

1

00 00

E, = ~ 21'+1 .i.-l._!l__.~/). _ """ 21'+1 -i •. !!'_.iJ'l! ••

r '::::"''''(''+1) 2:.siu.'f af{ .::::.. 1'.("+1) x a:1

1 1

00

JJ, == LJ (211 + 1). j •. ~;-. r..

1

00 00

"/ ~ 2" + 1 . 1 I. a I'. ~ 2 I' + t '. I: a I.l!.

JI.,= .--.,'- '-.--'_.- + /!'--., ,-.--,

~ 1'.(1'+1) 2:.810:1 a", .-... 1'.(1'+1) 2: iJ&

1 1

Broken and Reflected Waves

(SO)

12. If now we want to express the waves in the

interior of the sphere and those waves radiated outward

from it in the form of (13) and (14), we must take into account: first, that inside the sphere only the integral designated Iv of Eq. (10) can be used, because all other

-25-

particular integrals for x = 0, i.e. in the center of the

sphere, become infinite; second, that outside KvC-x) is to be used because only it has the factor

2."1 i r

t:-'~ == e 1.'

so the components of the electric and magnetic forces

contain time only in the combination:

.. '( r)

... ;l' ,., - ".1'

e

which is the characteristic of radiation coming from the sphere. The radiation inside the sphere can thus be

represented as follows:

DO

U _ ~ • I; .

• rl - L'/ 'Iv 'i"f ~.' 1

DO

lI"i - Yr (~, _1_v _. ~.~v + _q~. _I_/, a<u!) ,

• - --' 1<,(" + 1) Z, sill 7' a ~ ,',I" + 1) z iJ {}

1 '

-26-

(51)

The coefficients bv and qv' to one of which the factor

I is added only to make subsequent equations somewhat more elegant, can, in addition to the factors av and Pv appearing in the following expressions (52), be determined from the boundary conditions (7).

Outside in addition to the plane wave (50), which

appears on the wave, we have in addition the reflected radiation:

00

E ,-,. K,. ( - 101:) }J

:It"Q =- ,.:;;.;"",. J • -:r-. -. ,.,

t

00

s., = ~ ( a •• i _. ~,.:( - _x). ~ I',. + .2L_ . !F.r'!_~:1_ iJ 'Ur) •

__ ... (,. + 1) r. iJ tt , .. (,. + I) :11. sin ,'t 0 IJ'

t

00

" ~ A'.(-x)

JUr .. == ./ r p . __ . Il~ ,

._..'. x" +.

1

Go

u. .... ~ (_ .3..r..:i_. ~( - xl. iJ ~'r + --E1._. !i~J::_~. ~!..).

'P" ..... ".1 .. +11 :r rUt ".11'+1) %.siu:t off

I

Determination of the Coefficients

(52)

13. We have called the radius of the small sphere

p, the index of refraction of the medium (water) ,mO' the complex index of refraction of the material of the sphere

-27-

m.

Further we use a and S for the quantities x and x. for

a 1,

the special value p = p , and we use m' for the relative

index of refraction m/mO:

_ 211 • 'In" .f! 211 I!

a - ---, f-- - .. - At '

, 2:1.111." III ,1

,j = _OJ •• = -;/;o-·u = Tn -a ,

(53)

Finally we use the abbreviations:

I~(u) = .iJ~

I. ((j) = IJ. 11.'.(- u) = c.

(54)

Now we introduce into the boundary_ conditions:

(Ell + E~ .. )r = c = (Eo,). _!!' (E., + By, .. ). _ e - (l!',,')r ~ ~,

a . (lJJoJ + JllOa) , =e = fl· (M"d, _ CI u . (tlf., + J[., .. [, ,. e =- /J. (11/.,. i), -11'

the expressions (50), (52), and (51) and find, since two

pairs of these equations are identical, for av~ b

v~ Pv~ qv~

the following conditions:

(I) + 1) .• .i1: +- C: /. H,:

-_" .,.- a.--=I.--,

a·" • 1'1

- (2 II + 1). ir . A + U • C =-h . B ,

,. " y ,. "

(9 + 1) ." .A" + G" B, _

_ t' . I . -- p .. - -- = (1 • -,;. ,

a P« J" ~

(2 JI + 1). i· . A ' + P . C' = (I • B ' .

" r.. I" ,.

From this it follows that:

+ (2 + 1) . ~: • B •. fi - II,: . ~ •• II

a - " -I·· , I

• c:: . li •. (I - n: . 0 •. (I

(2 1)' .i1".B:.p-B ..... 1/.1J J

]I = - ,,+ . " . ----- - --- - --.

• O. ' B,: • fJ - B •. C: . IJ

(55)

-28-

Equations for Practical Calculations

14. The equations in (55) at first glance appear to

be so simple that it is difficult to discuss them. In the case of a metallic sphere, the problem is complicated by the

fact that a and of course Bv and B~ are complex quantities. But even in the case of nonabsorbing spheres, where a is

real, there remain two quantities that are always complex,

I

namely Cv and Cv·

If the spheres are not very large, it will always be

most comfortable to calculate with (27), (28), (29) and (30).

Accordingly we can set:

ar + 1

.-1. = 1.-s~.(2-;+i) '/.,

3" + I

B, = I-:-~-: :(:i-"+l) . s..

C =(_i),·I.S .. (21'-I).e_ia.(lt +iu.k)'

'" lI" ,.."

(56)

A ' = -~_"±:_!1.:_~ 'r.'

• 1.:1 .. (21' + 1) .'

B' =. (v + I), (1' •

• i-,T.~U;+-i)'9.,

C: - - I', (- i)' !: 3 .. (2,. .- .. 1.' , e> i .. ,(I, ' + i (( , It 'J

"",+1 "p.

The factors fv~ f' are rapidly converging power series
v
in a2; (lv' (l~ are the same series in a2; hv' kv,h~, k'
v
are finite power sums in a 2 thus all expressions which are
, relatively easy to calculate. They can be obtaiped directly from Eq. (27) to (30). All these sums fv' f~, etc., begin

-29-

with term 1, for a sphere whose diameter is small with

respect to the wave lengths of the light inside and outside, thus they all are very close to 1, and one obtains very

simple expressions for Av' A~ etc. If we substitute (56) into (55) and write for brevity:

. to' t. 9: h, + ill. k. go' (57)

u = e'G .-- , v = , IV = .-.

- ho'+i".k,' • I,' 9, • h,'+i".k,' "

(58)

then we find the following practical equations:

,,+ 1 ,,:!.+1 m't _ r

a = (- 1) .. -1 •. _._. . u .- - -_._- ._,

• /1 1'.3· .. l~/1-1)·· " .+1

III +. -_._. It'

• •

P,- (- t)'

,,+1 0'2.+1 1-1'

•. _-. ·ll . _ .. _--' _.-.

/1 l' . S' •. (:.I .. - 1)" r + 1

1 + --,,-' '''.

(59)

15. It is not difficult to evaluate approximately

how av varies with increasing particles size. For small Cl the f'l' g" etc are all rather precisely equal to 01 and

. "

consequently we can use the approximation equation:

a - a:!·+l."+_.~. C. l

• /1 1'. 3' .. (:! .. -:'I"ji ,

• m" - 1 J

c = --_. _._. _. (_ 1),0-1

, • " /1 + l' •

fIj + -- ..

(60)

Here Cu is a complex number whose absolute magnitude is not very far from 1 [8]. While av' as (60) shows, first increases with the (Zv + l)th power of a, for larger values

of a the av curve bends and never goes below a certain value. This can be seen if we take from (Z9a) and (l~) the values for I and K for very large arguments. Since a is real

v v

-30-

and S = S' - I·S" is complex, we have: }:(_«) .... e-io K'(-«) .... -i e=!»

" J. ., '

1. (a) .... sin (a - "2")' r: (al - cos (a -:- ~'~'!) ,

i fin i 1·fJ'-.._~) 1 po i I/I'_~)

/(3) - - -. 't: '1". '.! f ' (8) .... ' e 'C' \ ~

..... 2 ' J' ~ ~ , •

Thus if we substitute this into (55), then as a limiting value for very large a we find:

0.- (2,. + 1),1'., I

c.= ~. :~:, ,(sin (a - ";) + t . m". cos (a - ~2"))'

(61)

Here Cv is a complex quantity varying periodically with a, whose absolute value always remains in the vicinity of

1.

2v+1

Since the power a appears in (60), but in addition

the denominator increases rapidly with v, al must first

far exceed all other coefficients. But as al approaches its

.'

limit (61), a2 becomes significant, while all others still remain small. Later a3 joins the other two, etc. It can be seen from the equations for tv' gv etc., that Eq. (60) remains approximately valid for the subsequent coefficients.

'_

From this discussion it follows that, no matter how

large a, beginning with a given v, all coefficients

remain so small with respect to the first that they can

be completely neglected.

-31-

The radiation reflected from a small sphere essentially

comprises a finite number of partial waves~ but the

number of partial waves increases as the sphere becomes

larger.

This statement is first proven only for the "electrical

oscillations" of the particle which correspond to the coefficients a; it will also be confirmed for the "magnetic v

oscillations," which ihe incident wave excites, i.e. the

coefficients p , which we shall discuss in 17. n

In the numerical example shown below, I shall take a

diameter 2p = l80~~~ in order to show the optical properti~s

of colloidal metal solutions in more detail. This value

is within the range of microscopic resolution, thus it

corresponds to the extreme particle sizes of colloidal

solutions. EVen for the large diameter 2 p = 180 ~~,

beginning with v = 3 all higher "electrical particle

oscillations" are vanishingly small with respect to the

first two~ so we only need to caZculate al and a2.

On the other hand, e.g. in the rainbow problem, which can be considered with the method used here, a rather large

number of partial waves would have to be taken into account

and would result in considerable difficulties in calculation.

The first two coefficients are:

-32-

m"- ", m'~ + 2 "'. '

a = - -~- ·u6:u .

2 II ~

'111''! - t'.,

3

m' + -111, 2

1

(62)

If a becomes very small, then the u, v, w can be taken

equal to I and we find:

(63)

For small a, -naturally, a2 is vanishingly small with respect to aI' and only that first partial wave remains whose existense was first demonstrated theoretically by

Rayleigh, and which I shall consequently call the Rayleigh radiation for short.

16. We shall also consider briefly the special case in which the sphere consists of a completely conducting material. This case was first discussed by Thomson, and Ehrenhaft thought that the theory developed by Thomson could be applied to the optics of colloidal solutions.

'" ~

Although this does not follow for reasons discussed in

§l, the case does have a certain historical and theoretical

interest. We set:

m' 2 _ - i , 00 (cf. (3)),

and accordingly neglect in (55) the terms in a with respect

-33-

to those in 8. Then we find:

"._ (2/'+ I)'ir.~?

Then for small a:

. ?+1 ~.

a _ (_1).-1," + .. ~. II .eia. ..r __ •

• " 1', 3' , • (2,. - 1)1 II: + i,. . k:

For large a, (19) and (30a) give:

The absolute magnitude of this quantity oscillates regularly and periodically between zero and (2 v + 1).

The Rayleigh wave is of special interest here. The

exact value of al is:

(64)

For infinitely small a we find:

(65)

The characteristic difference between this equation and (63) is that the factor (m,2 - 1)/(m,2 + 2) appears which,

as we shall see, can be set approximately equal to 1 in

metals and which in any case varies severely with wave

length. Another very important difference is that in (65)

al is a purely real number, while the al calculated correctly from (63) has a substantial imaginary component.

-34-

This component, as we shall see, determines absorption in

the colloidal metal solutions.

17. We now discuss the "magnetic oscillations," Le.

the coefficients Pv' For the smallest particles equation (59) cannot be used, since (1 - vv) then approaches zero. According to (57), (29) and (30) we have:

1 - v - }-., (f.', !I - f. .!I ')

• /,,'. Or " ., " .. '

(66)

Here tv' as did the lv' I~ etc, indicates a rapidly converging power series in a2 and 82, which for small particle dimension is equal to 1. Furthermore we set:

(67)

and Eq. (59) gives for p : v

(68)

For very small diameters we can set approximately:

For large diameters, when the quantities Av' A~ etc. (54) are calculated according to (39a) and (19), (55) gives:

- 35-

lII'.Sill("- ... n) +i.C09(a-~)

Pv> (:! " + 1). i •. e!», 2 .r r=r:»> _ _2_. (70)

111 + 1

Thus the curve of Pv can be described just as was done in (IS) with the avo

For small particles Pv increases with increasing particle size approximately:

(71)

The absolute value of C' is very far from 1. v

For

larger values of a, the p curve bends and varies periodically v

for very large a with further increase in particle diameter:

1'.- (2/' + 1).C;.

C~ is a complex, periodically varying number whose

amplitude is not far from 1.

(72)

The same statements are valid for the magnetic partial oscillations as for the electrical, and indeed the v th

magnetic oscillation runs approximately parallel to the (v + l)th electrical oscillation.

By comparing Eqs. (59) and (70) ~ith (60) and (61) we see that the absolute value of Pv in general lies somewhat below that of av+l.

The first magnetic partial oscillation is, for example,

always of the same order of magnitude for all particle

- 36-

sizes, as is the second electrical.

Very smaLL partiales always radiate sidewards only the Rayleigh wave, for the very aoarse partiales in a aolloidal

solution the seaond eleatriaaZ and first magnetia partiaZ

osaiZLation are also of aonaern.

Thus I shall have to take into account in the numerical

example to be discussed below only three coefficients in

all, namely: aI' aZ and Pl' And indeed I shall calculate

PI for coarser particles according to the following equation:

C) 3 I - I',

PI::I - • ,U ,Ill --- • 1+ :!,c,

(73)

where ul' v~, WI are the same quantities as in (6Z). small particles the approximation equation

For

,,~

PI'" - -.(m! - 1).

Ib

(74)

is valid.

18. While the assumption of complete conductivity in the electrical partial oscillation lead to results very cornpar-able with theoretical predictions, for the magnetic

oscillations it produces completely deviant results. If in (55) we assume S infinitely large, we find:

I', -= _ (:! " + 1). ir . ~,r •

..

Thus according to (Z7) and (29) for small diameters:

- 37-

a value which is of the same order of magnitude as the value of av found under the same assumption.

If we assume that the spheres suspended in the medium

are perfect conductors~ we come to the opposite resuLt that the magnetic partiaL osciLLation of the vth order goes paraLLeL to the eLectricaL partiaL osciLLation of the

same ordinal number (instead of the next higher ordinal

number). In particuLar for very fine particles we thus come to the faLse resuLt that~besides the Rayleigh wave~

the first magnetic oscillation appears approximately in the

same order of magnitude.

Thomson, who first proved the last part of this theorem, has of course shown that the study of polarization of the light scattered laterally by colloidal metal solutions demonstrates that the result is erroneous. The characteristic

polarization phenomena observed by Ehrenhaft [9] and MUller [10] are no proof of the partial validity of the assumption of perfect conductivity. We shall see below that the

phenomena are clearly explained by the exact theory.

For the sake 9f the following numerical discussion, we

note the value of PI:

8 ; CI sill a - u ; cos n

PI =- . C 1 + ill'

(75)

For small ex:

-38-

(76)

If we compare (76) with (65), we see that the amplitude of the magnetic oscillation must be exactly'half'~hat of the electrical for very small, perfectly conducting particles.

Partial Waves

19. For the following it will be useful to have an idea of how intensity and oscillation direction of the individual part ial'wave's' (p ar t Lcu la r Ly 'th'e first) vary from point to point to point on a large sphere in whose

center is found the particle under discussion.

The components ~ and~. must not be taken into

, "

account, because they do not participate in energy transfer outside, according to the Poynting theorem. Thus on the surface of the sphere there are only tangential components. But now according to (13) and (14) we have for each partial

wave:

Btl • .1[" + Kip' 111.( .. 0 ,

i.e. the magnetic lines of force on the surface of the sphere are always normal to the electrical. Consequently to obtain a clear representation of the r add a't Lon. it is sufficient to define the electrical lines of force on the

surface of the sphere.

-39-

o

Fig. 3. First electrical partial oscillation

Fig. 4. First magnetic partial oscillation

Figures 3 - 10 show the electrical field lines on a

surface of the sphere surrounding the ~artic1e for the first four electrical and the first four magnetic particle oscillations. The plane of the diagram is the (1.3.) plane, i.e. the plane of oscillation of the light beam generating the waves. It is the only plane of symmetry involved,

and the planes behind the plane of the figure can be easily added to the hemispheres in front of those shown in the

figure, since the curves on both are congruent. For the

magnetic oscillations E = 0, so the lines describe closed,

1"

spherical curves, and on each of the two hemispheres at the

equator (8 = ~/2) there are v central points in which the force is zero and about which the field lines curve in v

different groups. In the electrical oscillations, on the

-40-

Fig. 5. Second electrical partial oscillation

Fig. 7. Third electrical partial oscillation

Fig. 9. Fourth electrical partial oscillation

-41 -

Fig. 6. Second magnetic partial oscillation

Fig. 8. Third magnetic partial oscillation

Fig. 10. Fourth magnetic partial oscillation

other hand, the lines of force lie on certain conical

surfaces which- all pass through v conical diameters lying in the plane of the drawing. The indicated curves are the intersections of these conical surfaces on the sphere.

These curves all spread towards the v poles cut through the v diameters. In reality the lines of force bend out of the

spherical surface in order to be included in the interior

or the exterior, depending on the phase of the oscillation, since they naturally can have neither beginning nor end (with the exception of the lines directly on the radiating particles).

We can now easily construct a-picture of the magnetic

lines of force. The 'figures of group I immediately show the magnetic lines of force of group II and vice versa, if they ~re rotated 90°, i.e. axes 1 and 2 are interchanged.

Diffuse Lateral Radiation

20. If we observe at an infinite distance away from

the particle, then we must substitute into (52)

as follows from Eq. (19). Furthermore:

I" ' () . - is

.ft,. -r _-I.e ,

thus:

- 4 2-

/

:!:dr

- --,T'

/. . -I' . i: e

~'I'j &I = + I. J Y'" = -l . ~-. . -.-

.... 1 r

co

. ~ (..u •. _., _~ ",. + I'. ,_ 1 __ • a:u!.) ,

..:::...' ,'.1.'+110'" ",(.'+1) ,ill.'J C"f'

I

E . )/ . i: e i.'

~.Q=-I.. o-a=-1'2;;' r

co

. ")' ( __ "» _ ... _1 .• O!'!, _ I'. . a i!!:..):

~ )"("+1) ~ill"t a., 1'.11'+1) ClQ

I

(77)

Here it is assumed that the light passing through the

solution is rectilinearally polarized, and that the electrical oscillation direction forms axis 1. According to (47):

-

Referring to the intensity of the transmitted Zight as

unity, the intensity of the diffusely scattered light is:

00

J... .l~' , I ,~(_ .. u ; _ 1 . iJ I'. _ 1",_ ,<1_ll\),',

.p ~n'."· :"::::"""("+11 "ill.'J i),p ".("+1) alt i

I

(78)

In these equations the straight lines indicate that

the complex quantities enclosed within them should be considered as absolute values. We recognize from (78), which

was already clear, that the intensity is inversely proportional to the square of the distance r. The two intensities

Ja and J~ indicate the components oscillating parallel to

-43-

the meridian (Je) and to the circle of latitude (J~).

Both components in general have a distinct phase difference which can be found by calculating the phases of the two complex quantities in Eq. (77). Thus if one observes at

an arbitrarily oblique angle to the beam passing through

the solution, one will obtain in general elliptically polarized light, assuming that the transmitted beam is rectilinearly polarized and that the suspended particles are so large that in addition to the (rectilinearly polarized for itself alone) Rayleigh radiation, higher partial waves are

also significant.

We must exclude however all directions that lie in the

two planes of symmetry ~ and ~ From the figures of §19 we can immediately conclude:

If the colloidal solution is illuminated with recti-

linearly polarized light and observed in a direction that is normal to its electrical oscillation, the lateraZly scattered light is rectilinearly polarized, and indeed its electrical oscillation direction is parallel to that of the beam passing through the solution. If the direction of

sight is unchanged and the oscillation direction· of

the transmitted beam is rotated about 90°, we again obtain rectiZinearly polarized light, but its osciilation 'direction

has also been rotated about 90° with respect to its previous

direction.

-44-

The first case will be characterized In our equation by 8 = n/2, the second by ~ = ±n/2.

The simpler the theorem above appears to be, the more important it may nevertheless turn out to be. Observations made by Steubing (cf. §l) of polarization of laterally scattered light showed that even if completely rectilinearly polarized light is used in this way. the lateral radiation will,still contain a very small component of unpolarized light in addition to the predominantly linearly polarized light. I am very inclined to conclude from this, that the particles suspended in'colloidal solutions cannot be sphere8~ even if the other optical properties may agree with this assumption.

The case where unpolarized light is used for illumination can be rapidly taken care of. We can imagine the unpolarized light divided simply into two incoherent beams of equal intensity that are rectilinearly polarized, the one so that its electrical oscillation direction is normal to the

plane determined by the bea~ direction and,the radius of vision, the other so that its electrical oscillation lies

in this plane. We then obtain in the radiated light two incoherent, mu t ua Ll y pc r pc n d i c u l a r 1 i nc a r l y polarized components, but in general of different intensity. This means that the radiated light is in part linearly polarized.

-45-

If an unpolarized light beam passes through the colloidal solution, the laterally scattered light is always in part linearly polarized (never ellipticaZly). And indeed the electrical oscillation direction of the polarized component is either normal to the plane determined by the direction

of the beam and the radius of vision, or it lies in this plane, depending on the size of the particle and the direction of sight.

This theorem is of course proven only under the assumption

of spherical particles, but it is not doubted that it is always valid in case of amorphous (thus not dichroic) colloidal solutions.

In solutions with very small particles only the first mentioned polarization direction is of course observed, but for coarse particles, as we shall see, the second may

also appear.

Intensity of the Light Scattered Normal to the Transmitted Beam

21. Since in general we observe with unpolarized light, we are only interested in the two major cases:

I. e = Tf./2, II. <t> = ±Tf/2. In both cases E<t> = Me = 0, thus

there remains only Je, which

we will distinguish in the

-46-

two cases as JI and JII. JI is the intensity of the radiation whose oscillation direction is normal to the

plane of vision, as we shall say briefly. JrI is the intensity of the beam oscillating in the plane of vision.

Initially (44) and (45) give for the two cases:

I iI' = ~. , sill If = " ,

. ~

\ J

I

(79)

ap u ( ) 1 at:" (1 2) 17'(" lI)( )

" = - lJ \V . __ ._... = -,). V,I-V. v.

a tJ " ai II if a ,'I " ,.

I I. rp = ± ;, sin ,'j. = ± u,

iJ I' 1'1 .• 17' ( 11 ' 'l 1 a 'U. -r II ( )

r, .. ' = + \' -1'-). V)-I: \'.'/" . '... = T • V •

o ~J -~ ,. ,. ~I U ... 0 If

(80)

This must be substituted into (78). First we want to

limit ourselves further to the case where we observe only

normal to the beam. Th us lye s et v = 0 an d now use Eqs, (40)

and (41). We find:

~.

l" 1/ " I:!") ((I., +1 )' I

J =._ .... __ . _-.!...+ .n\'_l)n+l, . ". _n_ ... +p-.

'4:t'r' :! ~ .,::n :!(,r+l) -".

\ -. ,

I

~ .

1'., ..... , (:! Ill, ( 1'" _\ )

J = - ...... ; ~"\'_l\,,+I. __ ~.lll!,.- .-.". ...

II 4 ~1'~ r'! I _:_, I 'J'~'J \ :!(T - 1 •

; 1 - !

(81)

These equations are written so that two coefficients of

the same order of magnitude appear in each summand (cf.

§17). In the numerical. example, as frequently mentioned,

I can limit myself to the coefficients aI' a2, Pl' I shall calculate with the following equations:

1'0 I 1/ I'

J =., __ !._I •

I "111 I '! I :! :

l f

(82)

". I '

J = A • <I," !'!.. I

11 4:02 r~ I :! I'

-47-

J11 now has a significant value only for the very largest particles, and even there JI predominates. Its value is

that of the Rayleigh radiation intensity, and it is at a right angle to the electrical oscillation producing it that

it has its maximum. Under other angles its intensity is:

and its total radiation is accordingly:

..

R J Q 2 J . S (~ I <L ~ 11 3 J.

". I'" It r. Sill II. I ,,- = - a . r . l'

o

Accordingly if the Rayleigh radiation is measured in the

principal direction, JI, we can find its total, i.e. the loss experienced by the light beam passing through it. Below we shall always give the quantity:

If = ~.),,'z.1 a, "

S7r I :! !

(83)

which at the same time is the measure for JI:

J. = _~_. R •

I 8:'l , .•

(84)

We shall see

in §26 how to find the total contribution

of radiation II.

-48-

Radiation of Many Particles

22. Previously we have always restricted ourselves to calculating the light diffusely scattered by a single particle. Now we want to consider the case where very many particles are finely distributed in the medium, all of the same diameter 2p and with the same optical

property m'. Then each individual particle is impacted not only by the directly transmitted beam, but also by the light scattered diffusely by the other spheres. The

effect of this second ether oscillation on the events in each sphere and the resulting change in the total lateral radiation has been studied experimentally for the smallest particles by Lorenz ; and Maxwell-Garnett has calCulated the consequences of the Lorenz theory for the optics of colloidal metal solutions (cf. §l). It is found there

that the suspended particles must be closely packed in order that the reaction of the lateral radiation should have a noticeable effect on the radiation itself.

We shall consider aqueous col~oidal metal solutions as optically infinitely dilute, and the total lateral irradiation will be calculated simply by mUltiplying the radiation that would be scattered diffusely by a single particle by the number of particles.

-49-

If any doubts remain about the validity of this assumption, they can be tested experimentally very easily by diluting

the solution a bit further.

If our assumption is correct,

the color of the solution will remain unchanged, absorption

in all parts of the spectrum being proportional to concen

tration.

If on the other hand the assumption is false)

the color must change, as Kirchner and Zsigmondy observed

in concentrated gold solutions in gelatin (cf. §l). So far

as I know, no one has yet succeeded in obtaining concentrated

aqueous solutions of this type. In any case I shall limit

the discussion to conventional dilute solutions.

In the following the concentration of the solution will

be given in cubic millimeters of metal per cubic millimeter

of water. If the number of particles in a cubic millimeter is N, their diameter Zp , thus the vo l umc V = 4np3/3, then

the concentration:

. ).,.

C= N.'·= .N.a3•

{; 11'

(8S)

On the other hand, according to (83) the total Rayleigh

radiation is:

thus:

R=l'~.C, (86)

where PI indicates a quantity that is independent of

-50-

concentration but dependent on particle size and wave length. We shall use two different equations for this quantity FI, depending on whether the particles are small or large:

(87) (88)

For this reason the first equation is especially

convenient for small particles, because, as we shall 3

see, al increases in proportion as a for very small spheres.

According to (62):

where ul' vI' wI take the value 1 for very small particle diameter. Thus:

~4,..8 ~! Ut'".! _ t't'

F = -- . / ·11 -------'

1 1" , I II.' + 2 Ii','

(89)

and for the smallest particles:

, 24 n" - 1111" - I I'

If =.. ./. .

1 l" , '11/' ~ + t ,

(90 )

This equation (90) is none other than the well-known Rayleigh

equation [11]. It says that, for constant concentration,

the radiation of a solution is the greater, the larger the particles. Indeed the, intensity of the 'radiation is

directly proportional to the particle volume. Furthermore

in general the short wave beams predominate, when the factor (m,2 _ 1)/(m,2 + 2)~ which depends on the optical properties

of the spheres~ does not vary much with the color of the

-51-

light.

But the latter presupposition does not apply in the case

of metals; the light which is diffusely dispersed by colloidal

metal solutions with very fine particles is generally not blue,

as is true of other turbid solutions, but has the color characteristic of the metal. If (m,2 - 1)/(m,2 + 2) is

constant, the well-known Rayleigh law holds, which says that -4

the radiation is proportional to AI for various colors,

all other conditions being equal.

Equation (88) allows us to apply the law for Rayleigh

radiation in the range for very coarse particles. As we

have seen (§IS), al for very large particles varies as a periodic function of particle diameter with constant

amplitude. Accordingly from (88) it follows that for

constant concentration Rayleigh radiation decreases with

increasing particle diameter, so long as the particles are

large. Indeed the Rayleigh radiation is finally inversely

. proportional to particle volume, with the exception of

periodic oscillation.

I. Given constant concentration and a very fine distri-

bution, the Rayleigh radiation increases as particle

diameter increases, approximately in proportion as the

particle volume; howevsr when the particles have become

quite large, it reaches a maximum and then decreases rapidly

if the distribution becomes increasingly coarser; weaker

and weaker maxima may still appear periodically ..

II. If the small spheres consist of perfectly conducting

-52-

or perfectly white material~ the particle diameter for which the radiation maximum occurs is proportional to the wave length~ and the maximum value of the radiation itself is inversely proportional to the wave length.

The second theorem, which follows directly from Eq. (88), is naturally not valid for colloidal metal solutions.

With some reservations, the appearance of this radiation maximum, which must also occur in turbid solutions of nonconductors (mastic, stearn, etc.), although at a different position than for conductors, can be considered as optical resonance. We must however keep in mind that this "optical resonance" is only accompanied hy very flat maxima, and that accordingly the energy spectrum of the diffuse radiation, even when all particles are exactly the same size, will exhibit only slight increases above an average value, if

the particles have become large. The radiated light will therefore be almost white in turbid material with nonconducting particles (stea, mastic), with only a very weakly evident color nuance. The colloidal metal solutions behave similarly; as we shall sec, for gold, for example, the yellow color is gen~rally evident, corresponding to the intrinsic yellow color.

-53-

Rayleigh Radiation of Gold Solutions

23. I now begin to calculate the optical properties

of a suspension of minute gold spheres in water, beginning

with the Rayleigh r~diation. I base my calculation on

the following numbers, which are found by drawing the flattest curve possible through the experimental data found by Hagen

and Rubens [12].

I have only changed the reflection

coe f f i c i en t in the violet-, which Hagen and Rubens have

certainly given as too small [13].

R 1'.1< R ".1<
i. (II. II. It) (11.11. It.) nn~. RU!!. . 11t m't l'
~:!tI n.:!!1:1 1,72 n.:I:!:! 1,7(1 l,iO - i. 1.7n n.oo - i. a,~o 313
450 U,:131 I,i:! U,:l:JtI IJ"l~ 1,73--i.l,i:! -0,UI7 - i. 3,:I~ 336
!lUO o,~70 :!,O7 O,4tiO :.?,O:! 1,10 - i . ~,()2 -1,(iO - i. 2,49 374
!l2!i O,(i 13 2,!!3 0,.9 - i, :!,:!3 -!?,·lr, - i. 1,:11::1 39:1
!l!lO 0,740 2,32 0,730 2,·l[j 0,!l7 - i . 2,45 -:~,20 - i .1,!l7 412
600 0,844 2,91 0,850 :1,96 0,38 - ,j • 2,:16 -4,8-1 -i.l,2G 450
:!l0 0,8S9 3,58 O,ASS :-1,54 O,~ I -i . a,54 -li,!)7 - i . 1,63 4lS7 In this table the first column shows the wave length of

the given color in vacuum, the two following columns give

the reflection and absorption coefficient from Hagen and

Rubens, the third and fourth columns the numbers that I

used for the calculation, m is the calculated complex index 2

of refraction of gold referred to vacuum, m' is the square

-54-

.'

of the complex index of refraction, referred to water (m,2 = = m2Im~) and finally AI is the wave length in water.

Further I shall introduce another practical abbreviation:

(91)

Then the Rayleigh radiation is calculated from (87) as:

(92)

The radiation for infinitely small particles is found by setting uI' vl~ w1 all equal to 1. For gold the following values are found for (°1)0 = (m'~ - l)/(m" + 2):

.I. _ 420 450 500 C,2C,

0,!l19 - i, 0,61b 0,602 - i , 0,666 0,801 - i ,1,180 1,380 - i , 1,4"'0

.I. - 550 600 6bO

1,925 - i , 1,211 1,8l:!0 - i. 0,391 1,Mb - i , 0,180

' .

thus in any case very different from what one would find for infinitely high conductivity. For perfect conductors we find from Eq. (65):

(a1)O = 1

completely independent of wave length. In general for

metals ~ has the form:

and i"lil = a'.' + a';' • In order to clearly demonstrate the difference with respect to perfect conductors, for which

~a11' .... 1 I shall cite the values of : 11 F' for gold:
, , 1
420 4bO bOO 525 550 600 650
0,i90 0,805 2,05 8,~4 5,11! 3,10 2,42
-55- ...

We can now picture the energy spectrum of the lateral

radiation from very small particles, since it is equal to

the product of concentration, particle volume and the quantity

The following table gives these values for

both perfectly conducting spheres and for gold spheres:

420 450 500 525
24 n:.l 7,76. 101• 5,85.10" a,SI.10'· 3,13.10'·
i:.- .1
24. n' 7,77. Ill'"
---y.-.- . i nil" 6,12.10'· ~,71 . 10'6 IU'':'.I(jlO
550 GOO 1;;',0 :H nl

--;:,-' 1 2,59. 10.6

24 . 111 .t

7" a, 13,37.10'·

r.ais • lO'~

G,iO. 10'·

:l,li . IO'~

Here I have chosen the millimeter as the unit of

length, and J will continue to do so below.

With the exception of the extreme blue and the violet~

the radiation of the gold spheres is stronger than that of the perfectZy conducting sphere would be.

One can thus perhaps say that the gold atoms resonate

on the light waves and best in the green-yellow. We must

therefore assume that in larger particles they are better coupled ~ith the vibrating ether than in small particles, because the radiation increas~s with particle volume.

-Su-

..

Fig. 11. 'Radiation of an infinitely fine suspension

Figure 11 shows a graphical ~epresentation of the

radiation of small gold particles and small perfectly

conducting particles.

(The dotted line shows the radiation

of the latter when the first magnetic vibration has been

added (cf. §18).). The last mentioned radiation precisely follows the Rayleigh law (proportional to ~,-4).

To find the radiation for coarser distributions, I had

,.,

to calculate u1' vI' WI' which consists of series iri a

(S2 = m,2.a2). I calculated for the values of a2 given in

the first column of the following Table I, and at the same

time I calculated (11 for the case of a perfect conductor for the same a2.

Then I calculated the radiation (87):

"4 s "

F = " . \'l • F ._~!._! = !~_ . a3 III 12

1 A,q :! fl' i ,i' . 1 .

The particle diameter belonging to each value of a2:

211=-~'lQ

, "

-57-

-.

Table I.

Values of the Coefficients 0. = a./2trl.

n" ,I Perfect
conductor .j:!1l 4:,0
0 1,00 0,579 - i.O,li7:, 0,ti02 - i , Il,tltitl
0,2 1,1).l - i. 0,065 O,·IH-I - , .. O,7!);) n.:,o5 . - .,: . 0,7·1:1
O,.l 1,04 - i.O,IHS O,3H - i.O.7:ltI O,:H;~ -i. 0.7:,7
0,6 o.sn I -- i. O,:H~ Il,!!:!·. - i . 0,6!1!1 O.:.!H ... -i. o,70li
O,!! O,H:II - £.11,-110 0,11:) - .:. U,6:1:! 0,1 :,Ii - i, (1,,;-10
1,0 11,1>38 - i . O,4:n O,O!l4 - i, ()':',;,!I 11.100 i.O,;,fil;
1,5 n,'IO:, - 1:. O,::C:" 0,0:18 -i. tl,4111 0,01:1 - i.II,.\Oti
2.0 O,:!f;:) - 1~. O,!!:,6 O,():!H -i.O,2:17 H,II'l1 - i . O,2U!l
:!,6 0,1!)() -_ i.0,11(; O,O:!tl - i. ()72:!:l Gold

0,~117 - i , 1,1S0 0,528 .. i.l,:ll:! 0,216 - i.I;2\1 11,042 .- i. I,O:!!I

-1l,O-17 - i. 0.84\1

-O,O:,r. - i . 0,715

-O,OH - i. (),4~()

-O,ou:. - i. 0,34H

n .. ld
1,:1 .. - - - .. -
;.:.0 "Oil li:,1l
0 1,!l2:, - 1:.1,211 l,fI~() - £.O,:!!H 1,545 - i . (I,I~O
0') I.SO:! --i. 2.0;-.0 2,1!10 - i. lI,ll77 1,!l20 - i.O,51;)
,~
0,4 O,9i1i - i.2,0-10 1,'flO - t, 1 ,~I'~ 1,767 - i. I,OSO
0,1; 0,0;)7 - i . 1,7l!1 0,R07 - i. i.nao 1,2:13 - i. 1,1!l~
O,S -0,107 - i. 1,:101l 0,11;(1 ... i. I.li I:! lI,ti7:1 - i.I,-t:l1
I,U -O,I:I·l -- i. 1,01·1 O,O!):, - t. I.:!Hl 0,:1;,:1 - i , 1,1!l1
1,:, - 0,07!! - i. 1l,1l:!7 O,Oll!1 - i. o, 7:! I 0,124 - ,: .0,7Itl
2,0 -0,022 - i,O,435 0,U32 - i. 0,479 0,096 - i.O,471
2,5 can be found from the following table:
at 1 - 420 45U 50O 525 Ij;'O 600 650
0,2 44,6 47,8 ;.:1,2 ;,;>,9 ss.e G4,0 6!1;3
0,4 6:1,1 6'i,li 'l :',:! 7:1,1 t:!:!,9 90,5 9::1,1
0,6 77,3 82,9 !)!!,~ 91l,tl )01,5 IIO,R 120,1
0,8 89,3 115,8 IOG,5 i u,s 117,~ 128,0 138,7
1,0 !l9,8 ror,o II!I,O 12.",0 131,0 143,0 15!i,0
1,& 1"" ., 131,1 1,1:',8 I [,:!, 1 11:0,[' 17[,,2 190,0
_.,-
,0 HI,I 151,3 IG~,3 rrc,s 18;,,3 ~U2,2 21\1,2
2,:> 157,7 16\1,1
Particle diameter in mm -58-

1,330 - i . 1,440 0,8:)0 - i. 1,823 O,2fla - i. 1,640

-0,028 - i J 1,347

-0,114 - i. 1,061

- 0, 126 - i , 0,855

-0,075 - i. 0,554

-0,029 - i. 0,395

Curves showing radiation as a function of particle

diameter are shown in Fig. 12.

(;0.:-- .. ----.,...---,-- .. -, ;::.----. :--1

I: I I 'I \ i I I

t--t--r - --1\ --j-( . \,-- : - --; _J,

I I I': v : I

so --'-"'-r--1--1 __ 1, [ \.- 1- .

~t I I !i;"''''', 1 !

-T-I--:t--~ --- \\-j--

I 1 1 ••• :-,..' 1"\'

~Of.- --,--;-j/-LI-; \\--1

j I i I ,'. I .J I \.. I -j __ L -li· .... 1 i\-'- ',.' I

,1( __ -t-_1..I. -If- i/-1 ... {. -. :\~\f";()

t I . -f. ;. \, \

.-- -II,' t. / I: ''\, : ",'\ \

I I ; /.... I'· I -. "',.\' \

, .. / . I '- i ' •. , ,

:!( __ , __ 1 __ ', :/i I· , .. _-." ~-"., , .

HI I I i I )_ .. l.... i -, ~ ('CJ --: --f·1 - ;t/. ! ;-"--: ~ ''_';'';

. ! I.~: .~I /~~~.::. ..... ,J ~ :-""

1Qr--·' - . ."1 .•. r'/... .. .,..._ ... ,.,fJIJ

Ili~ '~~;I' ii, -'<~;::-.,,;,)

I' hI, I'" · .. ·0

uf~I _ j_' .. _.. _J-J __ j'"

CJ 1.11 ','1 'oil h'O /f,(, 1,;."(1 ;." NIIJ ',.,fI,a .. W

Fig. 12. "Optical resonance" of small gold spheres

The abscissa shows particle diameter in mm, the ordinate

shows total radiation intensity R coming from a cubic millimeter of a solution with a concentration 10-6 (1 cbmm

gold per liter water), in thousandths of the intensity of

the light beam passing through a square millimeter.

The quantity PI can thus be found from the numhers on the ordinate by mUltiplying them hy 103, For comparison I have shown the corresponding curves (only the Rayleigh

radiation) in the same scale for spheres of a perfect

conductor. The diameters at which the radiation maximum

OCCurs are here proportional to wave length AI:

-59-

The maxima decrease with increasing wave length as l/A.

Fig. 13. "Optical resonance" of perfectly conducting spheres

With the exception of blue and violet, the radiation of

the gold particles is always considerably greater than that

of the perfectly conducting particles.

It is, as we said

above, as though the resonance of the gold atoms in the

yellow were added to the "resonance" of the particles.

The curve is highest at 600 ~~ (orange-yellow). The diameters

of the most intensely radiating particles are:

4:!0 4;)0 500 5:15 550 600 650
2 ~~I"x. 10;) 111 110 1UO !J(j 113 131
l' 11:1 336 3H :1\)3 H:! 4:;0 487
2 tl . ~lllx. 0,:136 0,331 (),:!~14 0,:!55 0,2:13 O,2!l2 0,269
l' Thus 2Pmax in general lies betweenA'/4 and A'/3.

The energy spectra of the radiations for given particle sizes can be found easily from the curves of Fig. 12.

The following numbers can be taken from the figure:

-60-

2(1 420 450 500 ;,:!b 5;,0 GOO GbO
20 0,262 0,1}19 (),338 0,5·16 O,G:17 0,299 0,139
40 2,10 1,1;0 2,72 ~,G2 ~)J~JU 2,H2 1,35
60 6,2 5,0 ::1,0 13,7 19,7 ]tJ!1 5,9
-,'.1
80 10,4 9,1 1·1,1 2:!,B 3i,H :13,0 14,7
100 1:1,2 12,0 li,2 ~7,1 42,:1 :l:J,[) 2!1,0
120 12,a 12,0 I i,1 ~fJ,O :w,u {.7,5 45,0
14U 10.0 10,~ 1;',0 :!U,t:! ~1,H 014,:) 4G,2
160 7,9 8,5 12,4 17,1 2~,5 113,3 un,s
180 6,3 ~" 1 0,·1 13,8 17,7 25,0 211,0
',-
The curves of Fig. 14 are constructed with these
numbers. We see then that: Small spherical gold particles must appear yellow-green

in the ultra microscope. The larger they become~ the more

their color changes towards yellow and red-yellow. Given

a constant concentration~ solutions whose particle diameter

lies between 100 and 140 ~~ radiate the most intensely; .

their particles send out primarily orange colored light.

The most intensely radiating solutions are consequently

brown in incident light.

Constructing the corresponding radiatioh curves from

the "resonance curve" of perfectly conducting spheres

clearly reveals that these color phenomena are due to the

special optical properties of gold.

I have demonstrated

this in Fig. 15 for particle sizes 2p ~ 100, 120, 140, 160 and 180 uu , These are already rather coarse turbidities.

While for 2p = 100 the curve increases quite sharply towards

violet, it flattens out increasingly with increasing particle

-61-

Fig. 14. Radiation of colloidal gold solutions

size and shows a very weak maximum which at 180 ~~ lies

at a wave length that is greater than 650 ~~. The radiated light from particles larger than 100 ~~ is in fact rather

white with a dark tone which, depending on particle size,

can be blue, green, yellow or red. According to Fig. 15

however, at 180 ~~ orange must be rather clearly pronounced; this is associated with the fact that we have limited the

problem to Raleigh radiation. If the subsequent partial

waves were to be taken into account, all curves and

particularly the last would be even more flattened, as an

easy calculation will show.

-62-

Fig. IS. Radiation from coarse, perfectly conducting spheres

Colors very similar to those associated with the imaginary

perfectly conducting particles will be obtained with color-

less particles. Using the equations that I used, one can

easily calculate the phenomena, for example, of colored

steam; the curves obtained will be similar to those in Fig. 15,

but with different particle sizes.

Polarization of Diffusely Scattered Light from Gold Solutions

24. Now we proceed to calculate coefficients aZ and Pl, For convenience we shall introduce new notations:

(93)

From (62), (73), and (74) it follows that:

,.1 ,"~1_ r

n_:l =' ---:; .. It'J ~ -''J - • ~>..... ,

1_ ... til + 't,t. Ir,!

(94 )

and for small values of a:

(95)

-63-

Here ul~ v1. w1 are the functions of 02 already used to calculate a1; "z: v2~ w2 are taken from Eq. (57) and (27) to (30).

From the values of 02 and PI shown in Tables II and III

we see that the second electrical and first magnetic partial

waves are significant with respect to the Rayleigh radiation

only for very coarse distribution. They appear most strongly

at approximately 550 ~~. If we calculate from (82) for 550 ~~ the ratio JII/JI for the radiation normal to the transmitted beam:

(96)

then we find:

t,'!

small

JII!J,

0,01(; . It'

lI,();'!) IiO /'.1'

0,642 1~5,a I'!'

One of the curves in Fig. 16 is constructed with these

numbers. Now as customary we can give the amount of

polarized light in the radiation in percent:

p = J,-JII .

J,+ JII

In our example:

U' 1)

2 ('

0,ti85 131 1'."

2

O ~.,

,--

-04-

Fig. 16. Polarization of light radiated under 90° at the 550 ~~ wave length

The quantity P is shown as the second curve in Fig. 16.

If the radiation of a colloidal gold solution is not completely polarized at an angle of 900 to the transmitted

beam, then it follows that the solution contains particles

whose diameter is much greater than 100 ~~.

Given a

particle size of 130 ~~, the polarization amounts to

approximately 90%.

Table II.

Values of

. a' 'I":'_~-~;o .. ,-"1 -- 4~-~' ~'! . 500 ~25 550 600 G50

klein {I a' X I a~ X I (11 X u' X I a! X i a' X ! ,,' X

0,0:.5 - i. 0,053 I 0,°';0-£.0,0521 O,OSi - t. O,Og I 0, 1 ~ 1- i .0,0:-;:) i 0, I ~!l- i. O,OG,' I , 0,131:1- i .0,021 [' 0,118 -i. 0,010

1 0,047-i.O,0;)!) I O,!)·19-·i.O,O;,8 i O,Oh3-i.O,IOI 0,IO:l-i.O,I:!4 i 0,1:'2-; 0,11,11 O,157-i.O,O~3 0,132-i.0,023

2 0,053-i.O,lll:Il 0,05;-i.O,119 I O,OaS-i.O,I:-IS: O,tJ52-i.O,:.!;,tl! O,IOK-i.u,!l22, 0,264-i.O,261 i, 0,2IiO-i.O,151

I' I!

2.r. 0,039-i.O,140: I: '!

-65-

Table III.

Values of

420

45()

50(1

550

uoo

650

,

{' a2 X (('! X (t:! :..;: U'J X : n:!;-:: 11' x I III x

klein I : I

i 0,033+,-0,107 0,034+i.O,lll 0,081+i.u.nS:l: ° 115+i.O,Ul>6; 0,1.10+i.O,052 0,I~l!i+i.O,O-l21 0,::!66+i 0,054

! 0,(145 + i. 0,065 0,04& + i. 0,01;1; U,OI):) + i. O,(J.I:! : O.(li:! ·Ii. II,O:!I . tl.lJ~ I + i. 0,02;, : 0,1 UO + i. 0,020 I 0, II t +i. 0,021 2 ; 0.0t;~+i.O,072 0,Otl5+i.O,Ui3) 0,018+i.O,U;>I l.o,()87+i.U.042 0,0!l4+i.o,o:n: O,II0+i.O,036. 0,123+i.O,OH 1,5 : 0,0(;5+i.O,071 :

Solutions with particles larger than 100 uu are now

those that radiate primarily yellow and red. We will see

below that they are always blue transparent.

Only the particles of blue goLd solutions radiate light

that is only slightly polarized at an angle of 90° with

respect to the transmitted beam.

This theorem can only be stated with a reservation,

since it is derived under the assumption of spherical

particles. There are in any case blue solutions that

radiate almost

completely polarized light at an angle

of 90°. But certainly one can say that there can be no red

solutions that do not give completely polarized light at

an angle of 90°.

If observations give a different result

(14), then it follows that the solutions were inhomogenenous,

that they contained many weakly radiating but strongly absorbing red particles and relatively few strongly radiating blue particles.

-66-

25. We now want to calculate the polarization of the

light at various angles with respect to the transmitted

beam. We shall call A the angle measured with respect to

the direction from which the beam comes.

In the two major

cases I and II (§2l):

I.

,'), - ".

- .,'

u ,~c I~OS j'

, ~ ,

J i:" ,-, r a " ' . '} !,

=-- .,-' /'" ,,(v.1J - (l-v~). fI')' + ,J. __ . fI '

11 4n-r" _ 1,',("+1)' • "'("+1) " •

1

Since we will deal only with the relative value JII:Jr we shall calculate only the quantities:

Furthermore we limit ourselves as always to the first terms.

We find

{ ::\ c= 1,1 - ('I --',) I")<"!

"'·1 I :! "1'·' I' ,

~" = I,ll' cos r - ,12·1:0,-;:.! i' -1- ~\I:.

(97)

Here 01,ll~,Vl are the quantities whose values can be taken

from Tables I, II and III.

By graphical interpolation we

find from these tables:

:! t' -~ 1 GO /'."

II, ,_. - O,OM - i , (l,ti:1

,I. = 0,1:' - i.o,:!:!

I', ~ o.on + i. (1,0:1

1 so ,,,,

-'1,(\:1' .: i, 0,47 O,I:! - i. 0,:10 0,1):1 'T' i. 11,1)3

for A = 550 uu

- () 7 -

If this is substituted into (97), then ~'+~Il gives the total

radiation as a function of the angle y, ~,-~" gives the excess of polarized light over unpolarized, P = (~,-~,,)/';j,+Su)

is the percent content of polarized light.

Since we are

dealing only with a relative value, I divided the individual

I f""b (<:"+~\

va ue s 0 '\~ Y ·S, ..J,"J.!"

; i.e. I referred everything to

the radiation taking place under an angle y = 90° as

unity.

The calculated values are:

:!(J = 101) I',u.
UO 20· 40" 60· 80·
Totn! } 0,64 0,67 0,75 0,88 0,97
radiation
Pol.1.rJzed 0 0;02 O,Of;
0,1 i 0,44
I' 0 O,U:I O,U9 O,:!l 0,46
101)" I:!O" 14U" 11'(j· 1.:i0·
Tota! } 1,00 L:I:, :!,Ol
radiation :!,";H :l,li
Polarized 0,8\)
0,~J7 0-') n,:!~) 0
,I ~
I' 0;;1; tl,,~ ll,:if; 1I,IIl U
:!~, :-: ':-iO ,",U.
0" :!O. ·H)" co u !'lOU
Totd J O,I:! 0,17
rod in t iOI 0,:11 u.ss 0,9;)
Polarized II
-U,O:! -0,10 -0,1;) +O,Ofl
l' () -0,11 -0;27 -(lt~:! +O,\U'
lOll" 1 :!ou Hllu lLiU· ISO·
Total } 1,1)·1
ra d i n t l uu 1,:1:1 e.iu :1,:10 3,85
Polar i .:er'
O,ti:! 1,IU O,U;, o ,:~:) 0
I' U,uO o.se U,:lII 0,10 0 90· 1,00 U,62 0,62

90· 1,00

0,31 0,31

In both cases the polarization maximum is thus shifted 90° towards increasing q and in one case (2p = 160 ~~)

towards y = 110°, in the other case (180 ~~) towards y = 120°. The fact that these are the angles is apparently due to the

-68-

Fig. 17. Radiation diagram of an infinitely small gold sphere

->

Fig. 18. Radiation diagram of a gold sphere 160 ~~ in diameter

->

Fig. 19. Radiation diagram of a gold sphere 180 ~~ diameter

special optical nature of gol~, since for other metals

generally smaller angles were found experimentally. The

negative sign in front of some numbers in the case of 2p =

= 180 ~~ means that here there is an excess of polarized

light that oscillates normal to the Rayleigh radiation.

-09-

The most striking of these numbers is however the high variability of the values with radiation angle. Given large particles, the radiation toward the side toward which

the transmitted beam is pointed is extr~ordinarily predominant. I have tried to make this behavior clear by means of several radiation diagrams in which I indicated the intensity of

the radiation by the length of the radius vectors emanating from the particle. The outer curves inFigs~ 17-20 cut off piece~ of the radii which are proportional to the intensity

of the total radiation, likewise the inner curves indicate

unpolarized radiation, the intermediate piece of the radius

vector is thus proportional to the polarized radiation.

-

Fig. 20. Radiation diagram of an imaginary, perfectly conducting, infinitely small sphere.

In conclusion we can say:

If the poZarization maximum of the ZateraZZy scattered Zight of ~ turbid soZution is shifted 90° towards increasing

-70-

y~this is a sure indication that the solution contains

very large particles. At the same time the lateral

radiation must then be extraordinarily much more intensive

towards the side towards which the exciting light beam

passes (y > 90°) than towards the other side (y < 90°).

For comparison I have also constructed the radiation

diagram for very fine particles and one for very fine

particles of an imaginary perfect conductor (Figs~ 17, 20).

It can be seen from the latter that the intensities are just

the reverse from the large gold particles. Should there still remain any doubt that the explanation of the polarization phenomena observed by Ehrenhaft and Muller is

correct, the question can be resolved very easily by comparing the intensities scattered under different angles;

this can probably be done with the naked eye.

Fig. 21. Polarization of the diffusely scattered light given infinitely fine particles

-71-

·,

Fig. 22. Polarization of the diffusely scattered light for 160 ~~ particles

Fig. 23. Polarization of the diffusely scattered light for 180 1J~ particles

roo ...

1/ f\
II
V 1\
V \
. __ v
r-,
o ".j. "". <lIP I/(J'': .IOfJ"- T.str l~tr }(;O- It.. 20~

-

Fig. 24. Polarization of the diffusely scatter~d light from imaginary. perfectly conducting, infinitely· small particles

Curves of P (Figs. 21 - 24) require no further explanation.

-72-

Absorption of Turbid Media

26. To calculate absorption we shall determine how much energy would be consumed by a particle if it alone were present. The absorption coefficients of the solution are

found by multiplying by the number of particles in a cubic

millimeter.

We imagine a spherical surface of radius r drawn about

the center of the sphere, and let x = 2~r/A' be a very large number. The current density of the energy flow through this spherical surface towards the outside is at

any moment:

We must now integrate this instantaneous energy flow over

one second to arrive at the intensity of the light. But

according to (2):

(i;" = ~1[Ei;.c2:ri"'],

[] means" real part of." If this is substituted, then

we find the desired time integral:

1& 1

J({f". '\)'1' - {f". ") o 1 . t! t = } [( A',). II,p) - (B,p' lll))] .

1= U

Here the parentheses mean that. the "scalar product" of the two complex factors is to be formed, i. e.

the product of the t~o absolute values with the

-73-

cosine of the phase difference. If:

E{) = E,; + i,E,~,

1',/, = 11,;. + t. J1,~,

where" E'e~ E"~ etc. indicate real quantities, then: ( i; . lJ y.) = 1:',;, J I.;. + I~':,', II,;~.

Since the same rules are valid for the scalar multi-

plication of two quantities as for conventional multiplication,

we shall simply omit the parentheses from the following

discussion. Where two complex quantities E and H are to

be multiplied, the scalar product will always be implied.

According to (4) we can replace the complex quantities He and H~ by i'Me and i'M~. The intensity of the normal components of the lights passing through the spherical

surface is then some factor times:

This factor is chosen so that the intensity of the beam

passing through the solution is equal to 1. We must

substitute for Ee~ M~ etc from (51) and (52) the values Ee E

e+ Eea~ M~ + M~a' Then the integral over the spherical

surface that represents the total amount of light passing

through the spherical surface towards the outside divides

into three components:

-74-

:I :!:,

1 = f j{E.). i J[., - E.,. i i11J). 7,2, sin o . a o. d (P,

It II

:r 2,.

11 = f j(E" •. ut; +N.?iJftpu - E'Pa,ijJ[{}-I~'.,..iJl.~a)r2 sin Itt dt'J-dc,.,

II (I

III = J j(l:""a' i Jf'l'u - B'f.' i J/" a) r2. sin o . ««, d 'f,

10 II

Component 1 contains the energy flow of the transmitted light beam. If it were not at all affected by the particle, it is immediately obvious that component I would be zero. Component III is a positive quantity, namely the total laterally radiated energy; thus it also gives the component of the absorption coefficient which is used to calculate

diffuse radiation. Component II is negative, representing

the total energy loss of the transmitted beam. Thus it gives

the total absorption coefficient which is determined both

by the diffuse radiation and the actual absorption of

radiant energy in the particle.

Since we assumed that z = 2rrr/A' is infinitely large

with respect to 1, then according to (29a) and (19) we

can set:

. I.' (.r) = cos (.r - ~;);

K. ( - .1:) = c - i z ,

K:(- r)=-i.c-iZ•

Furthermore if we introduce the following abbreviations:

-75-

A" =

;.' 2" + I, ("")

- . . I" - 1 . COS :r - --- ,

2" ". ( .. + I) 2

" "')

-- -I '

C =

"

;.'

- .. <:.: . e - i.~:

~n r, (I' -l- 1) ,

I'~ . f! - i z,

- 211 .... (,. + I)

j,'

then for the factors of the scalar products to be calculated

we find:

f' }' .... , .,1" + t'" rJ r,

~/} + ~"'" = ',.

_ r . ;1,'1

I

-

1

"; 11" + /I.,

sill ."

r

b'lI, iJ ;f' '

f' t: '1.,' .,1,,';- e,. I

,;,/ + :./., = /" ,

___ r ~ill 1'/

.

iJ i: -1- "~/I,, +-_lJ" a 'l;~

i1 'I' ...;:_. r ii /t '

1

- i . ~,", + ,II., .. )

,.,..,

~7.1:,. 1- (',.

I'

C')

ii I'... ,,' .1,. + /) .. rJ \l~t'

:-oi It ," ;,i " t _;_Y r ~ iJ:J '

1

_

I

on

'7.'.'1' !-('l

I'

a f',. aN

,'.1,.';' /I" a~.,

_;_'" r . sin t't . iJ If

I

i. : II,. +_ 11.,.,)

The light intensity itself is now calculated from:

(E" + };,.n).i{.ll.,. + .1/., ,,) - (N,I' + F, ,,}.(i(.I1, + ,If,),,)

:':0 00

= ;.,.; ')~. (A,:~ ~',\(fI" -i- (,,,j. (~f', ,;;. r: + .', . iJ I', .. i: ".,,)

_' ~ r' \ () :1 () ,'/ ~JII-," iJ 'f' 0 'f'

1. 1

oo 00

". ", (..t,.";_ (.',)' .-1 .. + /I,,) -i- I u,

- ~ I ,,' ,'{ .,

..- I":

~ ;.

C,.) II:,. -I- ",,)

. sill :/

'l'iJ_ I',. , i~ ll,. _ a/'., . il.'~,.)

t; ,', d '" iJ 'f d :, .

Wp ("::In nm.r integrate over the spherical surface as

follows:

-76-

2,J1':1 :!:tll

fl0[1. a P" . "'d·"" /"/'p 0(. ,0PI.) c

iJ.y·cT:t-·Sln •. • . If" = - .. "'a:; Slnd·iit; .dd.cirp

.j 0 U 0

Using Eq. (11) we find:

2,. "

ff(~ p'" ~~;,~. + " ~ ... ~,I_\ .~, PI'). sin «. u o . cl rt,

o :t 0 :t ~1U- .t o q' ii 'I' T

(\ U Zn ...

= 1'. (I''' 1) J J r, .. P" . sin o d o . d o ;

" U

Thus according to (46):

= 0, l' =l= f.L

I't .1,. + I)'

- 2 n . --2 ',.'+ 1'" l' = p..

We arrive at the same conclusion if we substitute

for Pv~ P~. Finally:

2~'1' :r

I ,r(~p,,· ,0 'PI' - ,o,p_ •. ~~!~). (J o . tl rp

" 0 & iJ (f 0 If' a :t

U 0

2."1 l)-=1I ;'I 'F-~"

= Ill',· °a~i-] . dcp - IIp''~iJ~r-] .d t't - O.

II .'=0 U '1'-0

The energy flow out of the spherical surface is accord-

.in g Ly :

00

2n.2}((.J.+ C,.).(B.+ (,',)+(..1.+ lJ,).(B + lJ))!"tJ~+ 11'_.

1 ' " r r 2,,+1

This expression is now decomposed into the above-mentioned

three parts:

00

"",', "i. (" + IF

II=2rr. ,/y(..1 +B).(C + D). __ ''_

,'" " l' Y 2" + 1

00

l'! "\' .-1 ", , , .

= 2;;' ~,.(- 1) . (Ie-'" X (a" - p_)e-")

1

-77-

00 00

III = <) ~~(('! + [J2) "t.( .. + I)" _ i.'" "-: (1.'+ 1'.,'

'" 7T. ... .... _., __ __.

1 :!JI+l :L~ ~ !!v+l

If the number of particles per cubic millimeter is N,

then II gives as the absorption coefficient of the solution

per millimeter:

(on)

l" - .,

k=N .. , .~.IIl .. >"(-l)"((/'-P.)'

~n -

1

(98)

According to III, radiation loss alone per millimeter

amounts to:

(99)

The symbol ~1lI () in (98) indicates that the imaginary

part of the complex number inside the parentheses is to be

taken. The vertical lines in Eq. (99) indicate that the absolute value of the quantities enclosed within them 1s

to be formed.

Aosorption of Colloidal Gold Solutions

27. We can limit ourselves to the coefficients a1, 82, Pl for the colloidal solutions. Furthermore as above we want to introduce the three quantities

-78-

Since further the concentration of the solution is

,>

".1. j.'] ti':I'! ,

C= X. r,

r=

we find:

(100)

For very small particle diameters, the quantity K varies

only slowly as the particles become larger.

Given a very fine distribution of the metal ~n

solution, the color is independent of particle size, absorp-

tion coefficient is simply proportional to the concentration

of the solution.

This theory can of course be valid only up to a certain lower limit on the smallness of the particles, since the

gold atoms behave differently in the optical sense than do small gold spheres. It would therefore be no doubt very

interesting to study the absorption of solutions with the

smallest amicroscopic particles and somehow to follow

op~ically the processes by which the gold particles are constructed from the atoms.

If the gold particles become greater than about 10 ~~, the theorem is no longer valid. I have taken the values

from Tables I, II, III, and constructed the curve

-79-

for each color which represents absorption as a function of

particle diameter (similar to the radiation curves in

Fig. 12).

From this diagram [ then read off the absorption

coefficients for the particle diameters listed in the

following tables. The numbers are given in the same units

as the radiated energy above.

Thus they signify the light

loss on a 1 mm path through a solution with concentration
10-6 (1 cbmm gold in 1 liter of wa t e r ) ln units per thousand.
The quantity K is thus found by multiplying by 103. Absorption of Colloidal Gold Solutions
~ (J ·I:!U : 45U n(lO n:!.~ !i.~O (iOO : 6:,0
.. -- _-._ "
0 ·.10,5 3i,·1 r,~), f: ti!J,() r,;J,:J Ifi,4 7,0
~l) ,"." I" " ,l~,:i 1'"., ';t;,I; (;I,l; I ~':' a.o
. -,' ,-,-
40 , ,II;,;) ·1:!,tJ tit;,1J :-i'i':l 7i,7 :!;',o 11,:1
60 .. ·I~,~ ·1:",·1 l, ~,.I ~!'{," ~ II;, I :11,:1 ifi,4
!!U " '·11:,:, .11,'; ti:!':) ~1.(1 !17,O li5,0 :!7, I
IPll ., 41.0 41,1) ~):!.:.! (;1>.:1 ~ ..• J ~5,U 44,5
. "'",-
l:!O ., :14,6 34.8 4:!,::i f) l,:! 61,0 77,5 58,7
140 " ~u,., ~~), .. ~t;-" ~) 41.U 4:-i,fi &.7,0 56,1
IGO ,. :!:>, i ~G,l :1(1,:; :lli,O 41,:! 4;" I 4;',6
IHO :!I,O :!L'! :!':I,O :1:1.0 Bi,t) .,- - :16,5
" "',' Thes~ numbers are shown graphically in Fig. 25 as the

absorption spectra for various particle sizes.

-RD-

Fig. 25. Absorption of colloidal gold solutions

Thus we see for fine distribution the well-known steep absorption maximum of the ruby red gold solution in the

green at about 525 ~~ wave length.

If particle size increases,

then for a constant concentration,absorption first increases

in the entire spectrum, color changes a little, gradually shifts towards the blue,and the maximum is always bent

towards the right. A severe change is first observed

when the particle diameter is about 100 ~~. For this

particle size1the solution is violet. At 120 and 140 we

-81-

already see the characteristic spectra of deep blue gold

solutions , at 160 that of an indigo blue , at 180 that of
a greenish blue. Thus in fact we obtain all the colors observed in colloidal gold solutions as particle size changes.

28. It is already clear that the color of gold solutions is primarily based on the absorption of light in the gold

particles themselves. They would never be observed with perfectly conducting particles. Here the light loss would

be entirely due to transverse light scattering alone, and

the absorption curves would therefore be identical with the

radiation curves shown earl Le r in Fig. IS. (if the next higher partial wave, thus the coefficients P2 of the corresponding losses were auued). These curves appear

quite different, like the absorption curves of gold solutions~

they would always represent only very flat colors, and for infinitely fine distribution the absorption would be zero.

For larger particle sizes the light loss through

lateral radiation also determines the color of the gold

solutions. Blue solutions are those that show a strong

red-yellow radiation.

It is in any case of interest to

calculate both summands of the absorption coefficient.

For a practical calculation we find from (99) the equations:

1<' '-: c .. I~!,

h" = :1 i.:' . u3 - ( \III~ + Yl.~ -f-

. 24 n' • ( 2

= -_ :, •... I. n1 +

I. .

,1 . 'II:., ~) I

!'I ...

~! • 0 .• 2). J

n "

(101)

-82-

Up to about 2p = 100 mm, KI is identical with the quantity Fl calculated in §23 (Fig. 14). For larger particle diameters, coefficients' PI and 03 are significant, although Rayleigh radiation continues to predominate up to 180 ~~.

I found the quanti ties Pl and 02 for coarser particles by

graphical interpolation from Table II and III, and calculated K' for the same particle sizes for which we already know K.

Then

K" = K - K'

(102)

is the measure for the fraction of the light lost in the particles themselves. The following table shows the calculated values of KII, again neglecting a factor of 103.

Coefficients of Pure Absorption

2 '! '! ":!O ; 4:)0 I !I00 -; --!I~~ .-:- . !I5-0 - . ; ---uoo __ l.__!_!10

- - '=;-~--;;=~~~---'" -'."'- - ~"~-~-1-"'-~'-~=-=-" 1 -

o i 40,5 I 57," 59,6! 69,0 I 55,5 16,' I 7,0

20Jl,u I 42,4 I 58,6 ! 61,9 i 76,1 i 61,0 18,2 7,9

40" I H,4 40,4 \ 63,:1 SI,!) ~1,8 22,1 9,9

60" il 42,6 40,4 MI,4 14,1 lli,4 :!5,O 10,5

80" '35,9 35,6: 48,3 58, 1 5~I,:! 32,0 12,'

100 " 27,-1' 28,7 34,\) 39,0 39,8 29,5 25,1>

120 " 21,1 22,0 25,2 25,6 24,5 19,7 13,6

140" 17,2 18,0 20,1 19,9 19,1 11,9 U,6

160 " 14,0 14,8 16,4 16,1 15,2 10,0 8,1

180." I 11 ,9 12,9 14,1 H,S 1S,3 8,8 6,5

These numbers are shown graphically in Fig. 26. These

pure absorptions curves always have their maximum in the

-83-

green between A = 525 and A = 550.

If we compare Fig. 26

with Fig. 14, we see no parallel between absorption and

irradiation. Thus for small particles, where irradiation

is still at a maximum in the green-yellow, the irradiation

increases very rapidly with increasing particle size, while

at the same time absorption exhibits a weak decrease.

Then while the irradiation maximum moves towards the right,

the pure absorption maximum also moves a little to the right,

although it remains always in the green. But in addition

to these relatively slight variations, we observe another,

which apparently does not depend on the special nature of gold and which finally predominates over all others: the

larger the particles become, the lower and flatter are

the curves. This is also ohvious; for when the particles

are to a certain extent thick, they are finally opaque for all colors, and the thicker they become, the smaller the

sum of the cross section casting a shadow.

If the bending effect is neglected, thus only geometric shadows are involved, the absorption coefficient would be kO = N"q, where at a concentration or 106:

.N=

thus:

iI . IO-U. -I !!

-84-

Fig. 26. "Pure absorption" in colloidal gold solutions

Then we can calculate for 2p = 180: kO = 8.3/1000,

for 2p = 160: kO = 9.4/1000.

I have likewise entered

these values of kO in Fig. 2S and ~b, naturally, as straight lines, because kO does not depend on COlOT. We see that diffraction makes absorption consideral)ly greater. Further-

more in the yellow and red by far the greatest part of the light is removed by the particles through reflection, while in the left half of the spectrum approximately half of the

lost light is absorbed in the particles.

The colors of colloidal gold solutions are explained

-85-

by the interaction of two ppopepties of goLd papticLes. The gold particles have in fact a vepy shapp maximum in the absorption capabiLity in the gpeen, and secondLy a maximum

in reflection capability in the yellow-ped. Vepy small paptieles reflect weakly and absopb strongly, consequently they make the solution puby ped. Large particles reflect strongly, at the same time the pure absopption curve for

them becomes lower and flatter, so they make the solution blue. The colop of the coarsest solution is alone determined by the property of the gold to reflect strongly the redyelLow part of the spect~um, this results in its less saturated blue-green color.

Of course this theorem can only be demonstrated under the assumption of spherical particles. But even for flake and rod shaped particles, we must distinguish similarly between absorption capability and reflection capability of the particles.

The comparison of the theoretical results found in this work with the measurement results will be left for Mr. Steubing. However it should be pointed out here that these measurements indicate that in fact the known optical properties of gold are retained for the ultramicroscopic particles and are completely sufficient to explain the optical properties of the colloidal SOlution, but on the

-86-

other hand the theory needs to be extended in order to explain everything, and should probably be developed further for ellipsoidal particles (flakes or rods).

Results

1. The problem of calculating the optical properties of turbid media has been solved under two simplifying

assumptions: first that the particles may be considered

,.

spherical, second, that the turbidity is infinite~y dilute

in the optical sense. The second assumption is certainly

appropriate for the conventional colloidal solutions.

2. The light irradiated by the small particles can be

calculated as a series of "partial waves, " and indeed

there are two groups of partial waves corresponrling to the

electrical and magnetic oscillations of the particles. Only a finite number of these partial waves needs to be considered for large particles, and the (v_l)th magnetic oscillation

is parallel to the vth electrical oscillation.

In colloidal solutions with very fine particles only the first electrical oscillation, which corresponds to the

"Rayle igh radiation" is of sign i ficance . For coarser

colloidal solutions, the second electrical and the first

magnetic oscillation are of significance.

-87-

<,

3. The assumption t hat the spheres ar~. p~F~e,~,!= ..... , conductors, which results in improper simplification of the equations, leads to another result, namely that the vth

,. .. th .. .. magnetic oscillation is parallel wIth the v .. electrIcal

oscillation. Accordingly even in the finest turbidities, the first magnetic oscillation must be noticeable in addition to the Rayleigh radiation, and this would give the Thomson theorem of the polarization maximum at 120°, Since the assumption is false, the result cannot be confirm~d

empirically.

4. If an unpolarized light beam is passed through the

turbid solution, the laterally irradiated light is wholly

or partially linearly

polarized (never elliptically

po La r i.ze d) .

Up to a particle size of about 100 ~~, the light scattered laterally by the gold spheres is almost solely

Ray Le i gh radiation, which has its polarization maximum (100 %) at 90°. If the particles become larger, then the contribution of the unp ol a ri z.ed light at 90° increases very

rapidly, and the polarization msximum shifts forward; at

Particle sizes 160 and 180 ~u it lies at 110Q and 120° .

At the same time the initial symmetry of the Rayleigh radi~

ation is disturbed by interference of the partial. waves

with each other; by far the greatest part of the diffuse

-88-

radiation goes towards the side to which the exciting light beam is directed. In the latter regard, the light reflected

by imaginary, very small, perfectly conducting spheres

would behave exactly opposite. The large gold spheres,

which exhibit this anomalous polarization behavior, have

a red-yellow diffuse radiation and color the solution blue. There can be no red particles that behave in this way.

6. Given a constant concentration, the diffuse

radiation of very fine turbidities is proportional to the

. volume of a particle. In coarser turbidities it increases with increasing particle size more siowly and finally

reaches a maximum whose position depends on the wave length.

In a turbidity of imaginary, perfectly conducting spheres,

the particle diameter corresponding to maximum radiation is

proportional to the wave length (O.324A'); for gold spheres

there is no law so simple, yet the particle diameters

concerned are always between A'/4 andA'/3.

The color of the diffusely scattered light, both for the

imaginary, perfectly conducting spheres and for completely white spheres, provided they are very small, is blue-violet,

corresponding to the Rayleigh law (radiation proportional -4

to A' ). Coarser spheres would send out an approximately

white light with a weak, dull hue, which depends on the

particle size. An optical resonance, which would vigorously

-89-

generate a given color, is excluded, at least for spherical particles, and it is not possible to explain the vigorous coloring of the colloidal solutions by means of resonance.

8. The diffuse radiation of gold particles is in general much stronger than that of equally large, perfectly conducting spheres. Furthermore, for the finest distribution, it exhibits a very sharp maximum in the green-yellow. Later

the radiation maximum increases especially high in the redyellow. To speak of resonance, we must first speak of a resonance of the gold atoms in the yellow, which is superimposed on the phenomenon to be expected in colorless conducting particles.

9. Infinitely fine turbidities inVOlving gold particles show (in sharpest contrast to what would be expected of perfectly conducting particles) a char~cteristic absorption that depends only on the quantity of suspended metal (concentration) and not on the degree of finess. It would be interesting to study experimentally the way in which

this may change in the case of minute particles which contain only a few atoms.

In. Generally ahsorption of colloidal gold solutions depends on two properties of the metallic gold: the absorp~ tion capability and reflection capability. Solutions in which the diffuse reflection opposes absorption show the

-90-

absorption maximum of the gold particles, which lies in the green; they are consequently ruby red. Solutions that show strong diffuse reflection are on the other hand bluetransparent, because gold reflects mainly the red-yellow light.

11. To complete the theory it is absolutely necessary to study the behavior of ellipsoidal particles.

Greifswald, Physikalisches Institut

Original article submitted January 7, 1908

-91-

FOOTNOTES

1. F. Ehrenhaft, Wiener Sitzungsber. IIa, 112, 181 (1903);

114, 1115 (1905). -

2. J. C. Maxwell-Garnett, Phil. Trans., 203, 385 (1904); 205, 237 (1906). F. Kirchner in his Leipzrg-nissertation ----also demonstrated the validity of the Lorenz equation for the refraction index of gelatin-silver emulsions (Ann. d. Phys., .!1_, 239 (1904)).

3. L. Lorenz, Wied. Ann., !.!_, 70 (1880).

4. R. Zsigmondy, Ann. d. Phys., ~, 573 (1904).

5. Cf. the discussion between F. Pockels and F. Ehrenhaft in Physik. Zeitschrift, ~, 152, 387, 460 (1904).

6. F. Ha s endh r l , Wiener Ber. IIa, 111, 1229 (1902).

7. Vol. I, p. 200 of the German translation.

8. In gold spheres, for example, the absolute magnitude of

Cl is between 0.9 and 2.5, depending on the color of the light.

9. F. Ehrenhaft, Ann. d. Phys., !.!_, 489 (1903).

10. E. Mii l Le r , Ann. d. Phys.,~, 1 (1907)~

11. Lord Rayleigh, Phil. Mag. (5), !Z_, 379 (1899).

12. E. Hagen and H. Rubens, Ann. d. Phys., ~, 1 and 432 (1903).

13. Cf. E. Hagen and H. Rubens, lac. cit., p. 453.

14. Cf. for example, E. Muller, Ann. d. Phys , , ~, 13, 16 (1907).

15. F. Ehrenhaft, Ann. d. Phys., .!..!..' 489 (1903).

16. E. Muller, Ann. d. Phys., ~, 1 (1907).

-92-

Sign up to vote on this title
UsefulNot useful